Sharing data can help reduce variability among providers.
There is little dispute that the United States is facing a catastrophic epidemic of opioid dependence with far-reaching health, economic and social consequences. Many forces are responsible for this epidemic, but overprescribing of opioids is certainly one of them. As physicians, we have the obligation and power to help fight this epidemic by ensuring that we use our prescription authority responsibly.
When we talk about tackling opioid overprescribing, some emergency physicians are quick to point out that EDs are responsible for a small fraction of all opioid prescriptions written. Indeed, data show that as of 2012, only 4.4% of prescription opioids were from emergency departments, and that share declined over time. However, the focus should be on initial opioid prescriptions.
In the ED, we often face the decision of whether to start an opioid-naïve patient on a course of opioids to treat their acute pain. Historically, a lot of the attention we paid to opioid safety in the ED was focused on avoiding prescribing opioids to “drug seekers,” i.e. patients who have already developed opioid dependence. For these patients, the horse is already out of the barn, and the priority should be initiating Medication-Assisted Treatment and referring for follow-up. When we’re talking about prevention, the greatest public health impact we can have is by avoiding unnecessary initial opioid prescriptions to opioid-naïve patients, and when we do prescribe opioids, following evidence-based best practices in how we prescribe.
Opioid prescribing habits among emergency physicians vary widely across states and among providers within the same department. A study covering 2011-2015 showed that the percent of opioid-naïve adult patients discharged from an ED with an isolated diagnosis of ankle sprain who received (and filled) an opioid prescription varied from 3% in North Dakota to 40% in Arkansas. And a study of over 300,000 opioid-naïve Medicare patients covering 2008 to 2011 showed that within an individual emergency department, the rate of opioid prescribing among clinicians (measured as what percent of discharged patients are prescribed an opioid) was 7% in the lowest quartile of prescribers, and 24% in the highest quartile of prescribers – within the same ED.
These variations in prescribing patterns can have long-term consequences for our patients. In the Medicare study, opioid-naïve patients assigned to a frequently-prescribing provider within an ED (a quasi-random process) were 30% more likely be taking opioids a full year later compared to patients assigned to an infrequently-prescribing provider.
In the ankle-sprain study, opioid-naïve patients who received a high-quantity prescription (greater than 225 morphine milligram equivalents [MME], equivalent to 30 tablets of oxycodone 5 mg) were 4.5 times more likely to continue using opioids long term (filling four or more additional opioid prescriptions in the subsequent six months) than patients who received a prescription for 75 MME, equivalent to 10 tablets of oxycodone 5 mg. CDC data from 1.3 million opioid-naïve patients similarly show a sharp increase in the probability of long-term opioid usage as an initial prescription exceeds three- to five-days.
Surely, you think, that level of variation can’t still exist in 2019. Our variability in prescribing must have narrowed by now. Well, if you haven’t looked at this data in your own department yet, you’ll be surprised. In my department, we had already implemented guidelines for safe opioid prescribing, required all providers to complete an accompanying online educational module, created individualized care plans for frequent ED patients and rolled out an Alternatives to Opioids (ALTO) program.
Despite all that, when we looked at our prescribing data last summer (see graph), we found that we mirrored the published trends: the rate of opioid prescribing (percent of discharged patients who were prescribed an opioid) varied among our physicians between 3% and 26%, and this variation was not explained by differences in patient population, acuity, shift distribution, etc. When we looked specifically at large-quantity prescriptions, i.e. greater than 15 tablets, we found that while most of our doctors never or rarely prescribed high quantities, two of our doctors wrote for greater than 15 tablets on the vast majority of their opioid prescriptions.
It’s all in the data
In the face of the opioid epidemic, there’s little excuse to have such wide variation among prescribers, but we know that we’re not alone. In addition to the published literature, when we’ve obtained the data from numerous different EDs, we’ve found the same variability over and over.
The good news is that this data is the key for driving change. Simply sharing this provider-level data with all of your department’s clinicians rapidly leads to significant narrowing of the spread as the outliers move closer toward the middle. We didn’t set any hard cutoffs or even a clear goal of what the “right” numbers are for prescribing frequency, but we did create a dashboard where all of our providers can see, unblinded, their data and that of their peers so they know where they stand. When I met with our most frequent prescriber to show him the data, he was shocked, as he honestly thought that he was a middle-of-the-road prescriber and had no idea his prescribing was so out of line with his peers.
Over the course of a few months, some of the sites that shared this data saw a decrease in total opioid prescriptions by up to 47%, largely driven by decreases among the previously high-prescribing outliers. When applied across tens of thousands of emergency department visits, these changes can have a significant public-health impact.
Certainly, there is a risk that the focus on curbing opioid use could lead to the undertreatment of acute pain. We’ve focused on non-opioid alternatives for pain treatment, but opioids are still the best treatment for some patients, and we’ve made clear that zero opioid use is not our goal. We’ve also closely followed our patient survey data, using a survey that our ED group mails to discharged patients after their visit, specifically focusing in on the questions that ask “how well was your pain controlled?” and about “concern for your comfort.”
Over the past year, looking at all of our doctors, we’ve found a slightly negative correlation between frequency of opioid prescriptions and scores on those questions, and we haven’t seen any decrease in those scores over time as we’ve reduced our opioid prescriptions. It’s important to continue to monitor this, as we want to make sure that the pendulum doesn’t swing too far, causing unnecessary suffering from unjustified withholding of pain medications.
When I informally surveyed a large group of ED directors from across the country, about half reported having looked at provider-level opioid prescription data in their EDs. If you’re in the other half, I strongly recommend working with your IT department to create a report. You may be shocked when you see the amount of variability among your providers, but you’ll be even more shocked when you see how effective sharing that data is at reducing that variability. The Emergency Department may represent only a small portion of the opioids prescribed in the country, but with the right data, we have the tools to have an out-sized impact in stemming the epidemic.
 Emergency Department Contribution to the Prescription Opioid Epidemic. Axeen, Sarah et al. Annals of Emergency Medicine, Volume 71, Issue 6, 659 – 667.e3.
 National Variation in Opioid Prescribing and Risk of Prolonged Use for Opioid-Naive Patients Treated in the Emergency Department for Ankle Sprains. Delgado, M. Kit et al. Annals of Emergency Medicine, Volume 72, Issue 4, 389 – 400.e1.
 Barnett ML, et al. Opioid-Prescribing Patterns of Emergency Physicians and Risk of Long-Term Use. N Engl J Med. 2017 Feb 16;376(7):663-673. doi: 10.1056/NEJMsa1610524.
 Shah A, Hayes CJ, Martin BC. Characteristics of Initial Prescription Episodes and Likelihood of Long-Term Opioid Use — United States, 2006–2015. MMWR Morb Mortal Wkly Rep 2017;66:265–269. DOI: http://dx.doi.org/10.15585/mmwr.mm6610a1External