Skip to content

Analytics key in fighting the opioid crisis

By Dennis Schmuland MD FAAFP, Chief Health Strategy Officer, Microsoft US Health & Life Sciences and Ryan Combs Samaritan Health Services VP Primary and Specialty Care on July 17, 2017

Filed under Health

Close up of doctor typing on laptop

Opioid abuse is a serious public health issue and a problem that affects almost every community and family in some way. What’s most alarming about the crisis is that opioid overdose deaths have recently risen to become the leading cause of death of Americans under the age of 50. In the last decade alone, the number of opioid overdose deaths has increased by 450 percent. That’s 44 people who die each day from prescription opioid overdose.

Worse, despite widespread awareness and national-to-local efforts to stem the crisis, it’s getting worse, not better. Between 2015 and 2016, overdose deaths in the U.S. rose an estimated 19 percent, the largest spike yet.

Local victories in this battle seem to be the exception rather than the rule. I came across one of those exceptions when I met with the VP of Primary & Specialty Care of Samaritan Health Services, Ryan Combs. Their community wide, data-driven approach is a bright spot in what feels like a losing battle. Below is the transcript of my interview with Ryan.

Schmuland: The prescription opioid battle has been going on for the last 10 years. What was the tipping point that compelled you to take action in the mid-Willamette Valley and central Oregon Coast communities?

Combs: We knew that Oregon was leading the nation in estimates of non-medical use of opioid prescriptions, but had little data to indicate how serious or widespread the problem was in our own backyard. So we felt that it was our responsibility to gauge the extent of the problem and, from there, formulate a plan and course of action to address the problem. But it was really the publication of the 2016 CDC opioid prescription guidelines that was the trigger that compelled us to act.

Patients with painful medical conditions need and expect safe, effective pain medication from the clinicians they trust. But, prior to the CDC guidelines, the lines between appropriate and inappropriate indications, dosing and duration of prescription opioids was left to the judgment of each physician. Once the CDC established these guidelines for starting, dosing, continuing, and stopping prescription opioids, the burning question became, “How can we easily and quickly dig into our own provider prescribing data, compare it against the CDC guidelines and show how prescribers trend against their own baseline and prescribing trends of our community clinicians?”

Schmuland: If you look deeper into the opioid epidemic it’s easy to see that there’ll never be a simple, one-size-fits-all solution. It’s a systemic problem with multifactorial determinants. There’s OxyContin being overprescribed, there’s cheap heroin now coming from Colombia and Mexico, and stronger lab-produced drugs from China and Mexico. How did you breakdown the problem into the areas that were in the scope of your reach and control?

Combs: You’re right about this being a systemic problem. The aspects of the problem that were in scope for us were the prescribers, the patients, and the community at large. So this is where we focused our attention and took a three-pronged approach: Build consensus agreement on an opioid prescribing policy, commission a task force, and make it easier for patients, families, and clinicians to tap helpful resources.

First, our Primary Care Medical Directors put their heads together to draft an opioid prescribing policy that laid out expected care guidelines for every prescriber, including regular urine drug testing for chronic opioid patients, weaning of patients above 90 milliequivalents, medication management agreements, and other expected behavioral health and risk screenings.

Opioid fact sheet infographicOnce we had an evidence-based policy in hand, Samaritan Health Services was able to spearhead an Opioid Taskforce across our three-county service area, bringing together representation from other provider organizations, county health departments, health plans, paramedics, and police and fire departments. The task force had three goals: achieve buy-in for this policy across the region, coordinate the efforts of every agency and organization dealing with the opioid epidemic, and connect patients and the public with resources to manage chronic pain and opioid addiction.

Connecting people to resources was the third prong. We put together a comprehensive website, Painwise.org, to empower patients and the community to learn more about the opioid epidemic and chronic pain management, including the ability to sign up for local classes, seminars, and even dependency programs. On that site, there’s also a section dedicated to equipping providers with the tools they need to do their part in this effort.

Schmuland: In what ways have you been able to leverage analytics to inform and accelerate your efforts?

Combs: We started out with static, patient-level Crystal reports pulled from our EHR to build lists of approximate daily doses for our patients. While the static reports were useful at the individual patient management level, our Operations Leadership and Medical Directors had no clear way to see prescribing trends more broadly and comparatively. We needed to be able to drill into provider-level data to assess the distribution of high-dose patients across the providers’ panels and to see how well individual providers were managing their chronic opioid patients. We leveraged Power BI for this, focusing on two different views. The first view showed provider prescribing trends, with slicers by hospital region, Samaritan department, and prescribing provider. We set as performance indicators the total number of patients prescribed to, total number of prescriptions written, and total milliequivalents prescribed, all trended over a rolling twelve-month period. Additionally, the Power BI report could be broken down by type of opioid drug prescribed as well as the age and sex, patients at different milliequivalent dosage levels and patients on Medication Management Agreements.

Schmuland: As a physician, I know that getting the attention of physicians and getting them all to agree on a policy is no small undertaking. How have the prescriber stakeholders responded to the program?

Combs: By and large the response to the program has been extremely positive. For example, the opioid prescribing policy developed by our Medical Directors was readily adopted as the standard of care for provider organizations across the region. Having hard data regarding opioid prescribing and patient dosage empowered our medical directors in crafting an opioid policy that reflected not only current medical recommendations, but also the reality of our regional landscape in relation to opioids. As a result, there were few, if any, provider complaints concerning the new opioid policy.

What providers really like about the policy was that it’s fact-based and broadly adopted across the region– which makes it much easier for clinicians to have the tough conversations with their patients about their overuse. Medical Directors also like the fact that they can track the decrease in opioid prescribing by provider and in the community, which illustrates that the policy is truly effective and worth the collective effort. Additionally, the ability to share this information with stakeholders external to Samaritan Health Services provides assurance that the Opioid Taskforce is affecting a reduction in opioid use across the broader community and has been worth their invested efforts.

Example data graphSchmuland: Have you seen any results to-date in the way of trends from your program?

Combs: Yes. Over the past year, we have seen a modest decline in total prescribing, and, while the total numbers of patients prescribed to remains relatively flat, the total numbers of tablets and milliequivalents being prescribed have trended downward. There has been a recent up-tick in the last month or so, but levels are still lower than they were previous to our concerted efforts to address the opioid epidemic in our region. With the completion of pending updates, Medical Directors will be able to quickly identify providers who may not be adhering to policy guidelines or simply have high-numbers of high-dose opioid patients. In this way Power BI provides executive-level data at a glance, identifies potential deficiencies, and drives conversations with providers to improve management of chronic opioid patients.

Schmuland: These reports look fantastic. Did you implement these Power BI reports on your own or did you work with a technology partner?

Example data graphCombs: Our Power BI work was handled entirely in-house. Especially on the operations side, we did consider third-party vendors that would supply stock dashboards for various business themes. However, there was considerable cost associated with those solutions. Strategically, we were concerned about what amounted to a black-box approach–there was so little transparency in the business rules and definitions employed by third-party solutions that we were concerned that we would end up with insufficient knowledge or ability to manage the underlying data structures or ETL processes.

In the meantime, it also became obvious that we needed to establish data warehousing capabilities to support not only executive dashboards, but also the ability to combine data from various sources for operational and financial analysis. A cost analysis showed that it would be cheaper to leverage data warehousing structures supplied by our EHR vendor and augment that warehouse with custom ETL from our other data sources. Additionally, the organization would be investing in skill sets and infrastructure for the future.

Schmuland: What were the factors that led you to select Power BI over the alternatives?

Combs: The ease of getting started with Power BI, as well as Power BI’s integration with the rest of our Microsoft SQL Server stack and with R was a major consideration in choosing a tool to layer on top of our warehouse. The ability to aggregate data from different sources within Power BI also meant that we could start building data dashboards and a data warehouse simultaneously, rather than requiring a fully mature enterprise-wide warehouse prior to the build and deployment of dashboards.

Schmuland: What advice do you have for your peers and colleagues that might be working on an opioid intervention program in their own region? Things they should do and things to avoid, for example.

Combs: I’d say as you begin to address the opioid crisis in your service areas, take a community-wide approach. The problem is just too big for any one organization to take on. It’s certainly easier to solely focus your efforts only on your specific organization. But because this is an epidemic that involves the entire community, you have to proactively engage every healthcare entity stakeholder that prescribes these medications and make sure everyone is handling these issues consistently. I would also suggest involving as much of the medical community early on as possible. At SHS, we initially looked to our primary care providers to help us address the issue, and their response was tremendous. But, in retrospect, we could have been more inclusive of our subspecialists, hospital based providers, and emergency rooms.

To learn more, check out our Care Coordination e-book here.

Useful Links

Contact Us