Next paper will be given by Tej Azad. He's a medical student at Stanford. Individual provider type is associated with opiate use in patients with newly diagnosed low back pain. This will be discussed by Dr. Ken Follett. Good morning. My name is Tej Azad and I'm a medical student at Stanford. I'd like to thank the selection committee for the opportunity to discuss our work on opioid prescribing patterns in patients with newly diagnosed low back pain. No disclosure of this talk. Low back pain is one of the most common reasons for seeking medical care in the United States. It's also expensive. In 2006, costs attributed to low back pain were greater than $100 billion with two-thirds of those costs associated with lost wages and reduced productivity. We also know that patients with low back pain present to a diversity of clinical settings and receive many different types of therapy, ranging from pharmacotherapy to physical therapy, epidural steroid injections and potentially surgery. This is a table from a 2017 systematic review in the annals of internal medicine. In the red box, you can see that for opiates, for either pain or functional outcomes, they found no evidence to support the use of opiates in this patient population. And taking the results of this systematic review with our observation of how many different types of providers that a patient with low back pain may be able to go see, we hypothesized that the initial type of provider who makes a diagnosis of low back pain may influence opioid prescription patterns in patients with newly diagnosed low back pain. So to investigate this, we identified patients with a new low back pain diagnosis in the market scan database. We then did a 12-month look back and we excluded any patients with a red flag diagnosis, things like foot drop, sepsis, a cancer diagnosis. We further excluded any patients with an opiate prescription in the six months prior to their diagnosis to identify, ideally, a population of opiate-naive patients. We then looked 12 months forward and eliminated any patients who had gaps in enrollment to require continuous follow-up, as well as if they developed a red flag diagnosis in the next 12 months. This left us with our final analytic cohort of newly diagnosed low back pain patients who are opiate-naive, about 500,000 patients. This patient cohort had a mean age of 49 years, was 56% female, and what I'd draw your attention to is the opiate prescribing statistics here, that even though these patients were naive in the six months leading up to their first diagnosis, 40% of them went on to receive at least one prescription, one opiate prescription, in the ensuing 12 months. And then 20% of these patients received an early prescription, defined as within two weeks of that first diagnosis. Four percent of patients met criteria for long-term use, which in this dataset was defined as six or more prescriptions in the 12 months of follow-up. This is our variable of diagnosis, with the Y axis being the percent of total low back pain diagnosis, and arrayed across the X axis, you can see the groups of different provider types. And as you might expect, the majority of patients are diagnosed in a primary care setting, about 50%. 20% were diagnosed by a specialist, this could be neurology, PM&R, neurosurgery. Another 20% were diagnosed in an acute care setting, like an emergency department, and the remainder were diagnosed by a non-MD provider. This ranged from advanced practice providers to chiropractors, acupuncturists. The first analysis we looked at was looking at the time to first opiate prescription. What you're seeing here is a Kaplan-Meier curve, with time following diagnosis on the X axis, and the Y axis being the increasing probability of an opiate prescription. And the two trends that we observed from this is that the yellow, black, and green curves, acute care, primary care, and specialty care, they rise quite quickly, quite early. That a lot of these opiate prescriptions are happening early after diagnosis, and the other observation that you can see is that the blue curve stands out, that patients who were first diagnosed by a non-MD provider appeared not only early on to perhaps be less likely to receive an opiate, but remained throughout the course of their follow-up, even after seeing other types of providers, seemed to be less likely to receive an opiate prescription. And so our takeaway from this was that that time to first opiate prescription varies with the setting of initial diagnosis. However, a limitation of this was not controlled for comorbidities or patient demographics. And to correct that, we built a multivariable logistic regression model that accounted for 31 comorbidities, age, sex, insurance type, receipt of surgery as well. And you can see the results of that here for early opiate prescription. The Y axis here is the risk of early opiate prescription, and arrayed across the X axis are the number of different provider types that we saw in our data. What really jumped out to us here is the acute care provider types. This is acute care hospitals, emergency medicine, and urgent care, that compared to our reference group, family practice, were significantly more likely to receive an early opiate prescription, again, within the first two weeks of diagnosis. This was relatively intuitive to us, that new low back pain diagnoses in an acute care setting may increase risk of an early opiate prescription. What we were really interested in, however, was long-term opiate use. Again, the definition for this is six or more prescriptions in the 12 months of follow-up. With family practice as our reference group, about a 2% risk of long-term use. What jumped out to us first was some of these specialty providers, specifically pain management and anesthesia, were significantly more likely, even after controlling for patient factors as well as demographics, to meet criteria for long-term use. We also observed that even though these patients who were first diagnosed by an acute care provider were more likely to receive an early opiate prescription, they were not necessarily then going on to meet criteria for long-term use. So our takeaway here was that that setting of initial low back pain diagnosis does appear to influence long-term opiate use. Our takeaways from this study was that presentation to an acute care setting increased the likelihood of receiving an early opiate prescription, that the influence of initial provider type persists well beyond the first visit, and that impacts the risk of long-term opiate use. And to take a sort of step back, we found that patients' starting point in the healthcare system may significantly impact where they end up and the type of care that they're offered. And this was really driving what we're looking at next, is this idea of care pathways. Are there pathways for patients who are newly diagnosed with low back pain that we can retain good clinical outcomes but minimize costs and potentially opiate utilization? Is there an ideal trajectory or are there less ideal trajectories for these patients? We're also going to start to look at benzodiazepine as well as opiate use, both through time, over years, and through space, looking at geography and geographic variation as well. Just in closing, I'd like to thank Dr. Ratliff, my mentor, as well as our team at Stanford, and also acknowledge my research support from the CSNS and the CNS. Thank you. Thank you for the opportunity to comment on this very interesting project. The research question, I found, is very novel, it's timely, and it's highly relevant in the current environment of concern about the opioid epidemic. Can you go back a slide, please? The primary strengths of the analysis relate to the very large data set which encompassed 75 million lives and led to the generation of 479,000 subjects. The primary caveats, however, are also related to the nature of large data sets. In general, large data sets can generate statistically significant but clinically insignificant results due to the sheer size of the sample population, and medical databases in general are subject to the vagaries of coding, and it's clear that hospital coders, clinic coders, speak a language that can sometimes be very different than the language that we as physicians use. The Truven market scan commercial database has some unique peculiarities of itself. Truven describes this market scan database as representing the American working population and their dependents. As such, it represents employees, spouses, and dependents of primarily large employers. This is a convenient sample in which subjects are selected because they're readily accessible. It's not a random sample, may contain biases, and may fail to generalize to other populations. The primary conclusion, which you just heard, is that for first-time medical encounters related to new diagnosis of low back pain, patients who are seeking care from emergency medicine and urgent care providers are more likely to get opioid early but less likely to use opioid late compared to patients whose first encounter for low back pain is with a pain medicine, anesthesia, or PM&R provider. This is a very interesting observation which I believe is an excellent starting point from which to understand what is different about patients who progress to long-term opioid use. Is it something about the patients, for instance, pain characteristics, etiology, psychosocial factors, or is it something different about the providers? For instance, pain specialists may have different referral patterns, they may be more willing to continue opioid, and ultimately, I think one of the significant outcomes of this project could be that it might help us target preventive measures most at risk for opioid use disorder. Thank you.

opioid prescribing patterns

low back pain

opiate-naive patients

acute care setting

long-term use