Editor's note: Dr. Juurlink is professor of medicine, pediatrics, and health policy at the University of Toronto, where he is also director of the Division of Clinical Pharmacology and Toxicology. We spoke with him about the opioid epidemic and strategies to address this growing patient safety concern.

Dr. Robert M. Wachter: What got you interested in opioid safety?

Dr. David Juurlink: What got me interested was a 2009 study led by a colleague of mine, Dr. Irfan Dhalla. We showed that in Ontario the prescribing of opioids had increased dramatically over the preceding decade or so. The increase appeared to be driven largely by the introduction of oxycodone onto our provincial formulary. But it was also accompanied by a sizeable increase in opioid-related deaths. That was our first foray into the study of opioid-related adverse events. Then I began to read a bit more about the pharmacology of opioids, and I realized, to my great surprise, that this practice I had engaged in quite liberally—using opioids for chronic pain—wasn't really supported by much evidence. I eventually learned that I'd probably harmed quite a number of patients with my well-intentioned prescribing.

RW: When do you think things started to change in terms of opioid prescribing patterns, and what were the original catalysts for those changes?

DJ: It's multifactorial, but it started in the mid to late 1990s. I'll give you some context. I was a pharmacist from 1990 to 1995, and I probably filled about 100,000 prescriptions. It was uncommon for patients who were getting prescriptions for morphine (which at the time was the strongest opioid we were dispensing) to have anything other than cancer. But in the mid to late 1990s, it became quite common for people to take even stronger opioids, specifically oxycodone, for chronic pain. The driver of that shift had to do with multiple things. One is that pain is common. Doctors see it in our offices all the time. We are rightly reluctant to use nonsteroidal anti-inflammatory drugs in certain patients. We've all been burned by their side effects. We have all seen acetaminophen not work very well for many types of chronic pain including osteoarthritis.

When we started to hear in the mid to late 1990s that we could use opioids more safely than we thought—we were principally worried about addiction—we were told that the risk of addiction was quite low (less than 1% was commonly floated) and that they worked well for chronic pain. Doctors were quite receptive to that message. Because if you were asked why you want to be a doctor, almost everybody says because they want to help people. So it's very gratifying to have a common problem—pain—and a tool to treat it in the form of strong opioids. Unfortunately the message that we could use opioids safely and effectively in the long term, without regard for doses, those messages really came from the companies that make these drugs, including Purdue Pharma and others.

RW: Is that what you think changed? Patients have chronic pain. It's often quite terrible. Physicians want to do something to make them better. Opioids have been around for forever. Did the pharmaceutical industry recognize a market opportunity, or was there something more going on?

DJ: It was largely the pharmaceuticalization of chronic pain. To be fair, there also were some compelling anecdotes that, for patients and doctors, reinforced the notion that you could prescribe opioids safely. With the pharmacology of these drugs, it makes a lot of sense that that would be the case. Let's say that I had a 55-year-old guy with chronic low back pain. For whatever reason, I was scared to put him on a nonsteroidal. But he was really debilitated by the pain. We've all seen patients who take oxycodone or some other opioid and in the short term feel a whole lot better. They can do things that they couldn't previously do. They often feel better mentally, not just the pain. So that can be quite gratifying for patients and for doctors. But what happens very quickly is people become physically dependent on opioids and they develop tolerance. I can unpack those concepts. These are sometimes confused and conflated with addiction, and that's entirely different.

Physical dependence is simply the normal adaptation to a central nervous system agonist, like opioids, benzodiazepines, or alcohol. But within a couple of days of taking opioids, your body comes to expect them to be there. So if I put a guy on opioids for a week or so, if he tries to stop them, he'll feel quite unwell. He'll develop opioid withdrawal, which is often accompanied by pain. So you can understand how patients quickly develop the impression that they need ongoing opioid therapy to stay well because the alternative is withdrawal. They don't necessarily perceive it as that. But they'll perceive it as "These drugs are treating my pain and I need to continue them."

The other phenomenon that happens is tolerance, which is simply a shift in the dose–response curve so that people need more and more of the drug to achieve the same effect. I'd be surprised if anybody who treats pain has not seen somebody who initially did well, then the drugs just didn't seem to work quite as well. The natural response, which we're taught to do by thought leaders, is to increase the opioid doses. We have millions of people in North America who have been on very high-dose opioids for exactly that reason. It's had a huge role to play in the genesis and propagation of what we now view as a more complex crisis.

RW: It sounds like a crucial part of the puzzle was this incorrect teaching that addiction is quite rare in patients who are given opioids. Where did that notion come from, and were data absent or was this truly misinformation?

DJ: It was misinformation. "Addiction is rare during opioid therapy" is a message that was parroted all over by key pain leaders. The most widely cited study has come to be known as the Porter and Jick study. So imagine you're at a fancy restaurant in 2000 and some specialist from New York is giving a talk about the role of opioids and he says, "Listen, addiction is rare and you can refer to the New England Journal of Medicine 1980 Porter and Jick. It happened in less than 1% of patients." Well, if you had the time and inclination to go to the New England Journal, you would find a letter to the editor that's five sentences long. It describes a database-drudging exercise at a Boston hospital that cannot possibly show the risk of addiction to be low. That letter only became available online on the journal's archives in 2010. There were a few other similar papers, but that one has been cited more than 600 times now, believe it or not. That single publication! About three-fourths of those citations have essentially reaffirmed the notion that addiction is a rare consequence of using opioids for chronic pain, which is not true. The best available evidence (which is not great) we have right now suggests that addiction occurs on the order of 8% to 12%. When you think about the fact that millions of people are on opioids, the notion that 10% of them might have iatrogenic addiction is really quite staggering.

RW: So were there actually good studies that showed that that number really is 10%, and these had been ignored in favor of these five lines in a letter to the editor?

DJ: No, the one I'm referring to is a study by Kevin Vowles. He summarizes the available evidence. One other problem that has a bearing on what we know about the long-term safety and effectiveness of opioids is that there aren't many studies. Most randomized controlled trials on this issue that establish the efficacy of opioids are short term; they last 6 or 8 weeks, or a little bit longer. Most compare opioids to a placebo or an active placebo (some low-dose benztropine or something like that). They very carefully select patients, people who don't have obvious risk factors for addiction; then they measure the pain scores at 6 or 8 or 12 weeks. That constitutes the bulk of the evidence for the effectiveness of opioids. So it's surprising, and all the companies really had to do to get the FDA and Health Canada, for example, to approve their drugs for chronic pain. They showed in those short-term studies that the drugs can do something and they appeared to be relatively safe. But what physicians have done is extrapolate those to people who would never have been in those studies, people who are on them for years, not months, at a time. And that's where we stand. There has never been a randomized trial to show that in the long term—at even a year, let alone 2 or 3—that opioids afford a more favorable benefit-to-harm ratio than nonsteroidal or some other class of active analgesics.

RW: You're painting a picture of what life was like in the mid to late 1990s—and the patient safety movement really begins in earnest in around 2000. It strikes me that this story then folds into that other stream. You have a situation where untreated pain becomes characterized as a harm. You have a new measurement schema where a pain is characterized as the fifth vital sign, and accrediting and transparency organizations begin to look at your performance on pain as a measure of the quality of care you're giving. How do you think that folds into this narrative about how things got out of hand with opioids?

DJ: It's hugely important. That phenomenon really affected the US more than Canada. In Canada, it has not been the case that my hospital's accreditation might be threatened by not measuring or assessing patients' pain at every clinical encounter. And my own license or remuneration, as sometimes is the case as I understand it in the US, has never been under threat by a survey of patients and did I do everything I could possibly do to relieve your pain. To me, it's unconscionable that those sorts of things existed in the US. But I think they had a large role to play in promoting what many people rightly see as the overprescribing of opioids for pain. Nobody will argue that doctors shouldn't try to treat pain, but the notion that opioids are a safe and effective way of doing that, as a blunt tool, is fallacious and extremely dangerous.

RW: Let's fast forward to today. In the space of only a few years, we have pivoted smartly on this issue and, in part through your research, have come to recognize the massiveness of the problem. What are the key elements of the solution?

DJ: One key element is to realize that opioids can cause pain patients an awful lot of harms. People sometimes call themselves legitimate pain patients. They have this line that separates a person with addiction from a pain patient. There isn't. But there certainly are people out there who will take their medications as prescribed and not double-doctor and not take extra doses and so on. It's important for patients and doctors to understand that some serious and often very subtle harms may befall those patients. Opioid-induced constipation is one of them. Sometimes that's a nuisance, but I've had people die under my care from that. Delirium, falls, fractures, subdurals, testosterone suppression, motor vehicle collisions. Physical dependence itself is a harm because it leads to self-perpetuating therapies for drugs that by their very nature wear off. So very often you have people who are just taking drugs because they have been taking them, not because they're getting good pain relief. The pain relief, if you look at the RCTs, clearly attenuates with time, even in the course of a few months of a study. I'm fairly confident in the assertion that millions of people in North America take opioids daily and think they need them. But what they in fact need is a very gradual, cautious, slow taper. Many of those people, especially those on the higher doses, will have their family report to you, once the taper has been effective, that the fog seems to have lifted and their old self is back.

I'll make one last point about these subtle harms. It's paradoxical and it's not widely appreciated, but these drugs can themselves cause pain. The phenomenon is called opioid-induced hyperalgesia. A typical story is somebody who started opioids for a bad hip or back. They progressed in doses, and when they get to higher doses they'll often look like somebody who has fibromyalgia. They have generalized pain everywhere. It's not recognized by the patient or even by the physician sometimes as an adverse effect of the drug. So you can imagine how that might leave patients and doctors in a Catch-22.

The first step is to realize that there are lots of harms other than addiction and overdose that can befall patients who are following the doctor's rules to the letter. If we were to step back and speak about what the overarching goals of drug therapy are, it's almost too facile to say, but the goal is to afford a patient more benefit than harm. This is why you don't prescribe antibiotics to somebody with a cold. Sure the risk is low, but the benefit is zero—so why on earth would you do it? What we see in patients with chronic pain, more often than we realize, are people who might at first have been being benefited more than harmed. But as time has gone on, as tolerance develops, and as doses have crept up, that pendulum can swing. So you can end up with a patient who is being harmed more than helped by their therapy and not even realize it.

So what can we do for those people? Have a difficult conversation. About half of Ontarians on chronic opioid therapy are taking more than 200 milligrams of morphine or equivalent per day. I don't think the numbers are quite that bad in the US. But if a patient like that comes into my care, the easiest thing you can do is just to keep them on it, I suppose. But the right thing to do is to have a difficult conversation. Say to the patient, "My job is to address your pain, but it's also to be mindful of the potential for some of your medicines to harm you—sometimes in ways that aren't really that apparent." If you talk to the patients and get their buy-in and engage them in, if needed, a very slow gradual taper, you will often have a patient who feels better as a result a few months down the road.

RW: You have pointed out that there is an initial period where the benefits may exceed the harms, and then this chronic phase kicks in where the harms may exceed the benefits. Does that mean that a short-term prescription can be a reasonable idea? Or once you're on that train the train is out of the station, so you shouldn't start it in the first place?

DJ: Let's focus on chronic pain because that's always where the most number of people reside, and there are people who say that you shouldn't use opioids for chronic pain. That's a bit dogmatic. It's fair to say that we use opioids for chronic pain a lot more than the evidence suggests we should. It's always an anecdote-based practice. So if somebody has really debilitating chronic pain, is it conceivable that you might improve their quality of life with a trial with therapy? I think it is. But it has to have a pretty clear plan in place a priori that you don't continue to escalate the doses, because all of these adverse effects are dose-related. So more careful patient selection and really resisting the urge to increase the dose, especially as time wears on.

If your 20–40 milligrams of morphine, or whatever the equivalent is per day, isn't cutting it anymore, maybe that's just a testament to the fact that this is not an opioid-responsive pain syndrome. Going up to 80 or 100 or 200 is the wrong thing to do, although we did it quite a lot. I did it myself. I don't want to suggest that we shouldn't be using opioids for chronic pain. The primary goal of medical practice is to relieve suffering and these drugs do have a role. The problem is that we cannot identify a priori which patients are going to be benefitted more than harmed. And it's always a bit of an experiment.

RW: I was thinking that as you said, "If you identify the right patient, it's reasonable to try it." It's hard for me to figure out who that is.

DJ: That's hard for everyone. There's no way of identifying who is not going to be harmed. We do have some tools that will allow us to identify people who are at particularly high risk of being harmed, at least from an addiction perspective. We know that people with a history of substance use disorders, people with histories of childhood trauma, sexual, physical, and so on—that's really asking for trouble. People who have ongoing alcohol use disorder, you're just inviting disaster when you commit that patient to try an opioid therapy. So we really can't know. That's why it's good to have an exit plan up front. "Hey listen, in the first few weeks of therapy if it's clear that you aren't a whole lot better, we're going to have to have an exit plan for tapering off these things."

RW: You've emphasized that we underappreciated and the literature underappreciated the risks of these medicines. Have we overestimated the risks of the alternatives?

DJ: Think about the alternatives we have, NSAIDs, acetaminophen, and the occasional gabapentin, pregabalin, and a smattering of tricyclics and so on. We've all seen the risks of those firsthand, and it's easy to impute causality when a person comes in with a massive upper GI bleed after a few months of naproxen, a little harder if they've had an MI on diclofenac or something like that. But my point is that what the patients on a nonsteroidal don't have is a drug or class of drugs that necessarily lends itself to ongoing therapy and to the perception of benefit that is specious. I cannot overemphasize this. The patient who has been on opioids for even a few weeks is physically dependent. They will really quickly come to realize that they feel crappy without their opioids. They feel better when they're on them.

We did a study a few years ago that took older people who were opioid naïve, went for minor surgeries (appendectomy, laparoscopic cholecystectomy, or cataract surgery), and then they were put on opioids. One year later, what percent of them were using opioids: 10%. These are people who weren't on opioids. They had an encounter with the health care system. They had an entirely well-intentioned postoperative script given to them. And 10% of them were using opioids a year later. That's a reflection of how these can be self-perpetuating. There is no conceivable way that the resident on orthopedics or ophthalmology intended this patient to be on opioids one year hence, but it happens.

RW: Do you think the health care system is set up to manage this problem? You've talked about how part of the solution is sitting down, having a hard conversation, and really taking the time to go through the risks, alternatives, etc. Immediately what pops into mind is a 13-minute primary care clinic visit when the patient has 5 other problems.

DJ: Yeah, we're not even close to prepared. That doesn't even consider the people who have opioid use disorders, which is a separate and growing category of patients who need a different sort of intervention. Our system lends itself to a quick encounter, and there's a societal expectation that there's a pill for every problem. You can't sleep, there's a pill for that. You can't stay awake, there's a pill for that, too. You have pain, here's a pill. Insurers are quite happy to pay for pills. They're less happy to pay for physiotherapy, occupational therapy, or cognitive behavioral therapy for certain patients. So our practice models really facilitate the treatment of pain with medication, which would be excellent if we had drugs that were safe and effective for that problem. We just don't. The way that we interact and treat patients and the forces that influence our daily practice tend to perpetuate this problem.

RW: What do you think the lessons are from this sad story for the health care system and for the field of patient safety specifically?

DJ: Something that's too good to be true probably is. The notion that we could take compounds that were derived from the opium poppy, chemically modified, and give those to people in a way that was safe and effective in the long term—seems overly optimistic to me. It was a message that we were conditioned by nature to hear because of how much we want to help people. That's one lesson. There's another lesson that we haven't touched on. I think it actually harms some patients with pain. As much as I don't endorse the use of opioids, particularly at high doses as a treatment for chronic pain, in the US in particular we have seen doctors in some instances abandoning patients and in others just rapidly reducing their doses to comply with, for example, the CDC published guidelines and imposed dose thresholds. I know that patients were harmed because of doctors who were eager to avoid a knock on the door from the DEA.

So when we have people on high-dose opioids like this, the wrong thing to do is to cut their doses off quickly. We need to step back and always deal with the patient in front of us and ask ourselves what makes sense now. Sometimes what makes sense is doing what you're doing. But other times, what makes sense is a cautious, perspicacious reassessment of the big picture. If your goal as a doctor is to help more than harm, we need to seriously rethink how we prescribe opioids.

RW: You could easily see this as a story of our inability to balance these complex forces appropriately with the idea of pain as a fifth vital sign and the pressures that created to aggressively treat pain. Obviously that plus marketing plus lack of an appreciation of the risk caused us to go too far in one direction. It sounds like you worry that we may go too far in the other direction.

DJ: I think in some instances we have. I've often been lumped into the anti-opioid camp. There's a pro-opioid camp and an anti-opioid camp. The reality is that there's some kind of balance there. There is a role for the use of opioids in the treatment of chronic pain. We just don't know what it is. We don't have studies to tell us who is a good candidate and who is a bad candidate, although we can identify some of the latter intuitively. Not to sound cliché, but it's more art than science. But if you're going to dabble in the art, you have to have a full appreciation of the benefits and harms. It's only after a 2-decade experiment on the population that we are starting to really get a good handle on the harms. And there are a lot more than we thought.