The opioid crisis in the United States continues to reverberate across society. While a holistic solution remains elusive, health plans and pharmacy benefit managers are evolving their use of data analytics and technology to improve prevention and treatment among their members and in their communities.
The opioid crisis in the United States has had a widespread impact on all aspects of society. The epidemic extends across multiple delivery points in the health care ecosystem, with no single entity capable of implementing a complete solution. While a holistic approach involving the entire ecosystem is likely required, health plans and pharmacy benefit managers (PBMs) have an opportunity to curb opioid misuse among their members and in their communities—by leveraging data and technology to improve prevention and treatment. To help identify some potential strategies, the Deloitte
Subscribe to receive related health care content
Read the related "My Take" by Sarah Thomas, Managing Director of the Deloitte Center for Health Solutions
Learn more about the Deloitte Center for Health Solutions
Explore additional perspectives and services related to fighting the opioid crisis
Center for Health Solutions interviewed 35 clinical, pharmacy, data analytics, and policy leaders from health plans and PBMs across the country. We found that a growing number of health plans and PBMs are taking a data-driven, evidence-based approach to help change patient and physician behavior. We expect emerging technologies to play an increasing role in supporting these endeavors in the battle against the opioid epidemic.
According to our major findings, potential strategies for health plans and PBMs include:
Despite limitations in the availability and completeness of data, many health plans and PBMs are using the information and evidence they do have to develop leading practices. Many of their programs focusing on educating consumers and clinicians about prescription guidelines have successfully curbed prescription rates. Although there are clear health care and societal savings associated with helping someone overcome an addiction, the best strategies for long-term success are not always clear.1 For many health plans, addressing opioid misuse among their members and in their communities is a key strategy for improving health outcomes. PBM stakeholders told us that taking on this challenge is a critical part of their mission and an important opportunity to mitigate future financial and reputational risk. Although prescription rates for opioids have decreased in recent years, many industry observers agree that the rates of addiction, overdose, and death will likely get worse in the coming years before they start to improve.2
This complex, multifaceted problem calls for systematic solutions across the health care system.
Prescription opioids can be effective in treating pain. However, sometimes, whether taken alone or in combination with other drugs, they can lead to abuse, addiction, and in some instances, life-threatening adverse events.3 Opioid misuse is frequently front-page news, and overdoses and opioid-related mortality rates have been increasing for more than a decade. There are many reasons for this, but they go beyond the scope of this report. Briefly, they include an intense focus on pain management, marketing campaigns, quality metrics that are built around pain management, a gap in the understanding of chronic pain and addiction pathways, and (until recently) a lack of clear guidance on the appropriate use of opioids.4 Health plans and PBMs, like other stakeholders in the health care ecosystem, are working to address this crisis.
Although the United States has just 5 percent of the world’s population, we consume 99 percent of the world’s hydrocodone and 81 percent of its oxycodone. According to some estimates, opioid use in the United States is 30 times higher than is medically necessary given the size of the population.5 Here’s how the opioid epidemic is affecting employers and hospitals:
Substance abuse is not new to the US health care landscape, but the demographics for opioid misuse are different. In addition to urban areas, the opioid crisis is affecting many suburban and rural communities. Workers’ compensation claims were an early indicator of the emerging opioid crisis, more so than health insurance claims or other signals. A 2012 report by Lockton Companies concluded that prescription opioids were the top driver of indemnity losses related to workers’ compensation claims.8 The report noted that prescription opioid abuse stemming from the management of chronic pain had the most damaging impact on claims.
Many of our health plan and PBM interviewees said that in the early years of the epidemic, their organizations had focused on monitoring prescribing data to avoid FWA. From this data, they began to realize over the past five years that patients were often treated with too much pain medication or for longer than recommended. This typically happened after routine surgery or dental procedures. As a result, ER admissions for opioid overdoses and opioid-related deaths were rising. Health plans and PBMs realized that the health care system needed to reframe the issue. The substance abuse and addiction models of the past need to be rethought. They also knew they had to look at a broader data set beyond what is traditionally used to target FWA.
As health plans began to review their data, descriptive analytics helped them bring the issues into sharper focus. They started to identify some of the metrics they wanted to target, such as prescription rates, total morphine milligram equivalents (MME), and duration of therapy. Health plans also began to look at which diagnoses were being treated with opioids (where the medical and pharmacy data could be linked). They were then able to identify certain clinicians who had higher opioid prescribing rates than their peers. The CDC’s 2016 guidelines on opioid prescribing were helpful in directing clinicians to appropriate, evidence-based opioid-prescribing guidelines.9
The substance abuse and addiction models of the past need to be rethought.
OUD is hard to generalize and can affect anyone regardless of geography, age, income, education, and other factors. However, many health plans and PBMs began to see patterns emerge in the data that could help identify certain factors that might increase a patient’s risk of becoming dependent on or addicted to opioids.
As health plans analyzed their data, they could see who was on a long-acting opioid (not recommended as a first-line drug for acute conditions) versus a short-acting opioid, which might be more appropriate. In 2018, a large study showed that patients who were prescribed long-acting opioids were 2.5 times more likely to suffer an accidental overdose than those who were prescribed short-acting formulations.12 Given the amount of inappropriate prescribing, many health plans added an edit to their claims systems, which directs physicians to prescribe certain medications as a first-line therapy, with certain quantity limitations.
Placing limits on the duration of prescriptions was one of the most common early intervention strategies used by health plans to curtail opioid abuse. The 2016 CDC guidelines were helpful in creating their messaging. But our interviewees emphasized that it is not enough to limit supply. In some cases, that strategy could lead to unintended consequences. Research shows that limiting opioid prescriptions for those already taking opioids can sometimes open the door to heroin use. Heroin can be cheaper than prescription opioids.13
Research shows that limiting opioid prescriptions for those already taking opioids can sometimes open the door to heroin use.
Our respondents recognized that some physicians have complex patients who are dealing with chronic pain, multiple comorbidities, possibly mental health issues, and other complicating factors. Interviewees cited a need for more effective risk-assessment and decision-support tools that are either online or integrated into electronic medical records (EMRs) to guide clinicians trying to manage these more complex patients. The number of technology-based tools that can help clinicians is increasing. Table 1 provides examples and use cases that are starting to hit the market and could be deployed more widely in the future. The challenge is integrating these technologies into the clinical workflow, having dynamic tools that can be updated as patients’ conditions evolve, and integrating the technology with the practice of other clinicians treating the patient.
We knew we could not just put a quantity limit on prescribing across the board. It’s not just about setting rules and changing coverage. There are patients out there who are dealing with complex chronic pain issues, and we need to make sure those patients with both substance use disorder and chronic pain are getting effective treatment. We have invested in a number of web-based tools and supports for the care team and the patient. And we are piloting many different comprehensive approaches to pain management. We are also working to get information on effective MAT programs and providers to our primary care physicians.
-Doug Nemecek, MD, chief medical officer, Behavioral Health, Cigna Corporation
Novel analytic techniques can help health plans and PBMs leverage their data to identify people at high risk for opioid misuse. Many health plans and other health care stakeholders are in the early stages of creating predictive-modeling tools. As they learn more about the drivers and predictors of opioid misuse and overuse, they are likely to strengthen these efforts.
The duration of opioid treatment following surgery was the strongest predictor of opioid abuse among commercial health plan enrollees with no history of misuse or ongoing opioid use, according to a 2018 British Medical Journal study.14 Each additional week of opioid treatment increased the risk of dependence or overdose by nearly 20 percent. Each additional refill increased the risk by 44 percent.
Some health plans are segmenting their populations into different risk categories, and are studying how people move in and out of the categories. This can improve a health plan’s ability to identify people who might be at risk early on, and offer interventions at the appropriate time. For example, the lowest-risk groups might include members who have no SUD claims and no use of nicotine or other substances. The highest-risk categories might be members who have SUD diagnoses from ER visits or other hospital visits, but are not currently seeking treatment. Health plans and other stakeholders that use predictive modeling can continue to refine their models and test out different interventions to continue to learn what works.
In addition to data analytics and predictive modeling, emerging technologies can also be used to help fight opioid dependency and addiction. As seen in table 1, many health plans and PBMs are already leveraging some technologies, and could begin using others to reduce opioid misuse in their populations and communities.
In 10 years, I believe in-person office visits for behavioral health will be the exception, and telehealth visits will be the rule. It will be the standard of care as the technology continues to improve and people become more at ease with it. For the patient, it is often more convenient, private, and they may be more comfortable in their home or in a familiar setting.
— James Schuster, MD, chief medical officer, Medicaid, Special Needs and Behavioral Services and VP, Behavioral Integration, UPMC Insurance Division
Although data analytics is typically an essential part of health plan and PBM strategies to target opioid misuse, our interviewees cited many limitations and barriers to using it to its full potential. These include lack of interoperability and siloed systems, limitations of PDMPs, and privacy regulations that need to be modernized. Increased collaboration across the ecosystem could help resolve some of these challenges, but policy changes may also be necessary.
PDMPs collect data from pharmacies to track the prescriptions for controlled substances that patients have filled. When physicians or dentists check their state’s PDMP database, they can look for worrisome patterns of opioid prescriptions. From there, they can deny or change a prescription or educate the patient about other options or addiction treatment.
Studies show that PDMPs can help change prescribing patterns and reduce possible harm from opioids.26 But PDMPs have limitations:
Efforts are under way to improve PDMPs. The Prescription Monitoring Program Training and Technical Assistance Center at Brandeis University, for example, provides a wide range of services and supports to PDMPs, federal partners, and other stakeholders to build on and improve the effectiveness of PDMPs.29
Our interviewees, all of whom rely on data to shape their strategies, expressed a desire for additional data, from a variety of sources beyond their purview. For example, some organizations have only medical data. Pharmacists and PBMs are often limited to pharmacy data. Some organizations have both, but lack behavioral health data, toxicology data, or data from EMRs. This data could help them refine and improve their data analytics and predictive-modeling tools. It could also help them develop better tools to help clinicians deal with complex patients.
Health plans and health systems are also starting to collect more data around the social determinants of health (SDOH)—factors outside the health care system that influence health and may drive addiction and OUD. Such SDOH data may include information on a person’s environment, income, access to healthy food, or transportation barriers that make it difficult to access care. Our interviewees agreed that SDOH data is sparse and not easily accessible at the point of care. A recent Deloitte report shows that only one-third of hospitals are integrating any kind of SDOH data in the EMR. And, just because the data is there, it does not mean that the care team is accessing it. Another finding of the report was that the SDOH data residing in the EMR often gets buried in unstructured data such as social work notes, and is not accessed or used by the clinician.30
As discussed earlier, even if clinicians have access to these additional data sources, there are frequently challenges with integrating them in the workflow, ensuring that the data is updated, and integrating the data with the practices of other clinicians treating the patient (see table 1 for early solutions that may be more widely deployed). This was reinforced by speakers at a March 2018 hearing held by the US Senate Health, Education, Labor, and Pensions Committee on the opioid crisis, who highlighted challenges with data-sharing.31 Some of the expert witnesses stressed the importance of having a holistic view of the member, patient, or person in the community. Sanket Shah, a health informatics professor at the University of Illinois, recommended ways in which federal agencies could integrate multiple data sources at the local and state level. Having a centralized data repository could help further advance predictive analytics and identify high-risk individuals earlier. He also asked the committee to consider supporting the Prescription Drug Monitoring Act of 2017, which would require states that receive federal grant funding to establish a PDMP to enable data-sharing with other states. The act would also fund a data-sharing hub to serve as a central repository.
Having a centralized data repository could help further advance predictive analytics and identify high-risk individuals earlier.
Some of our interviewees noted that the lack of data-sharing was not solely a technical problem. There are also cultural issues related to data-sharing, as well as an inherent risk. Any entity that handles health care data needs to ensure that it has systems in place to secure it and keep it private. Additionally, a 46-year-old privacy law (42 CFR Part 2) protects the disclosure of SUD diagnosis or treatment information to avoid deterring patients from seeking care and then potentially facing stigma from employers, insurers, housing, child custody, and other situations. The law requires the patient’s consent whenever this information is disclosed. Many stakeholders view some elements of this law to be a barrier to improving behavioral and physical health care integration.32
To encourage more data-sharing, the federal government and states need to have appropriate regulations in place to navigate security and patient privacy concerns and the tricky issue of who owns the data.
Interviewees expressed a desire for more data to help them advance their analytics and predictive-modeling tools. With various stakeholders having different information about a member, patient, or person in the community, it is difficult for any stakeholder to get a holistic view of the person and intervene early.
Our interviewees discussed the stigma around OUD and the role health plans and PBMs can play in fostering the acceptance of OUD as a chronic condition. Comparing the use of medication for OUD with treatments for type 2 diabetes, many of our interviewees illustrated how attitudes toward OUD were different. While most health care professionals accept that a patient diagnosed with type 2 diabetes should continue with a medication therapy that works, this is not always the case with OUD. As discussed in the sidebar “What tools are health plans and PBMs using to target opioid misuse and abuse?” there is substantial evidence that MAT can reduce overdose deaths, illicit drug use, criminal activity, and risky behaviors, and help improve health status, functionality, and quality of life. Despite the evidence, some people believe that at a certain point, a person should not be treated with medication for OUD. Instead, the goal is abstention from any opioid. Many of our interviewees said they felt the need to educate their own staff on the evidence, as well as their network providers and community. Other suggested strategies include increasing access to MAT through benefit design and improving the integration of behavioral and physical health.
Another challenge in efforts to support treatment is the shortage of supply of SUD treatment facilities and staff. Waitlists persist in almost every state.33 A 2017 Health Affairs analysis reveals significant gaps in access to MAT across the United States34 (see Deloitte’s interactive tool for more information). During his first few weeks as secretary of the US Department of Health and Human Services (HHS), Alex Azar, touting MAT as a critical tool in the fight against opioid misuse, stated that only one-third of SUD treatment programs offer MAT. In response, the administration aims to raise this number through initiatives, such as new guidance from the US Food and Drug Administration, and by encouraging new studies around MAT.35
Deloitte developed an interactive map using data to highlight where MAT treatment gaps persist in the United States. The map shows the distribution of facilities that offer various medications for MAT at the county and state levels, and how that distribution corresponds to some measures of the opioid crisis, such as opioid prescription rates, age-adjusted death rates of people aged 15 years and older, and key demographic data. Key findings from the tool include:
Other efforts by health plans to increase access to MAT include:
Many interviewees also said that the lack of standardized quality outcome measures for SUD and opioid treatment can make it challenging to secure high-quality treatment for members. Cigna is working with the American Society of Addiction Medicine in partnership with researchers to validate treatment outcome measures. As part of the collaborative, a team from Cigna shared two years of medical, pharmacy, and behavioral health data. Many health plans are also working with Shatterproof, a nonprofit organization focused on addiction, to develop treatment quality measures.36
Moving forward, health plans and PBMs want to evolve value-based payment models along with the rest of the health care system. Value-based payments for behavioral health issues, including opioid disorders and SUDs in general, have traditionally lagged behind medical and surgical conditions. One major challenge is the lack of standardized quality measures. Some health plans are piloting programs around prevention and treatment, and are advocating for policy changes to address the limitations in data.
Despite the lack of standardized quality measures, some health plans are developing value-based care payment models around opioid use. California, Rhode Island, and Vermont have different programs for opioid dependency in their Medicaid programs.37
This epidemic is multifactorial. Segments of the US are experiencing sixfold greater levels of opioid prescribing, leading to increased rates of addiction. We know there are pockets of the country where the overdose and death rates are profound. And we know there are other, complex issues at play, including social determinants of health, hopelessness in the face of high rates of unemployment, and the emotional and financial stress people who have family members struggling with opioid use disorder often face. We need more research to understand and address all of these issues.
— Hal Paz, MD, MS, executive vice president and CMO, Aetna
The health care ecosystem has several stakeholders that engage individuals at various points in the care delivery chain. These include prescribing clinicians, PBMs, retail pharmacies, health plans, behavioral health providers, employer plan sponsors, and policymakers. Previous Deloitte research “Fighting the opioid crisis: An ecosystem approach to a wicked problem,” has framed an ecosystem approach to address the opioid problem that includes public health stakeholders, economic and workforce development, the criminal justice system, and child welfare and other social services, in addition to the health care system.
Traditionally, as each of these stakeholders has wrestled with how to prevent and manage OUDs, they have developed a series of solutions at different points along the chain. Health plans recognize that a more comprehensive approach is required to address the opioid epidemic. This involves different types of interventions across the three main pillars of the prescription life cycle: identification, prevention, and treatment and recovery (see figure 3).
Ecosystemwide approaches might also require policy considerations. Our interviewees noted that their organizations are supporting policy changes, including modernizing privacy regulations, developing policies that aim to increase the number of mental health and behavioral specialists, and mandating electronic prescribing of controlled substances.
Interviewees also discussed the need for more research on chronic pain as well as effective non-opioid pain-management therapies. A 2018 study showed that prescription opioids were not more effective than over-the-counter drugs or other non-opioids in treating chronic hip or knee pain, for example.38 To date, there is limited evidence about the use of acupuncture, spinal manipulation, and yoga to treat different kinds of pain.39 Because they do not involve potentially addictive medications, stakeholders are interested in exploring these alternative methods. In 2018, Ohio’s Medicaid program became the first in the Midwest to cover acupuncture for the management of low-back pain and migraines.40
In addition to understanding pain and pain relief better, interviewees also called for more research on SUD and addiction. They want to know more about the factors and comorbidities that put people at risk. They also discussed the importance of creating a deeper understanding of the SDOH, specifically factors related to the opioid crisis, some of which include unemployment and corresponding feelings of isolation, and family members with addiction.41
So many people and organizations are addressing the opioid problem with different strategies. It’s a fragmented approach, and we risk competing amongst ourselves for attention and resources. What we need is more collaboration between the health plans, health systems, and community partners. At Geisinger, we are working with several different community nonprofits in Pennsylvania and in a few other states. We are working with pharmacy schools at universities. We are sharing data from multiple sources and developing a coordinated strategic plan for the county.
— Perry Meadows, MD, medical director for government programs, Geisinger Health Plan
The common thread weaving together the various interventions for health plans and PBMs is the sophisticated application of analytics to unlock powerful insights from the data. Reducing barriers that prevent data-sharing across stakeholders can allow various players to apply more holistic solutions. While policy levers can play a role in the future of the opioid crisis, the organizations we interviewed understand the critical need for more collaboration across the ecosystem. They recognize the need to join forces with other health plans, the clinician community, and local community partners. They are reaching across the health care sector, creating and joining coalitions and collaborations in their communities to break through the silos. By combining efforts, they can work through these complex issues, move the research agenda forward, and bridge the gaps in data.