A Healthesystems publication

Winter 2016

Thinking Outside the Claim - Using Multidimensional Data to Estimate Potential Opioid Misuse

FAST FOCUS: Development of a predictive model that incorporates multidimensional data can enable us to better anticipate the treatment needs, considerations and challenges of specific patient populations – enhancing our ability to monitor and intervene in a claim with unprecedented insight.

Understanding population data can help payers better understand the specific treatment needs and potential challenges of patients within a population (see “How to Treat a Lady: Drug Therapy Risks for Women in Workers’ Comp”). But acknowledging these data only scrapes the surface of possibilities. The natural evolution is to then synthesize and construct data in a way that not only helps illuminate these needs and challenges, but enables us to anticipate them before they occur.

Gathering and analyzing large-scale, multidimensional data over a period of time lends itself to the construction of a predictive model on which patient management decisions can be made that impact future outcomes.

GETTING AHEAD OF OPIOIDS

One example of how population data can be used to impact future claims outcomes in a significant way is by the identifying the risk for potential opioid misuse. Considering four of the top ten drugs in workers’ compensation are opioid products comprising 15% of total pharmacy spend,1 getting out in front of opioids would serve to dramatically impact the cost of a claim as well as the overall health of the injured worker.

MULTIDIMENSIONAL VIEW OF RISK

There already exists a number of established risk factors that prescribers should consider when weighing the risks and benefits of opioid medication in a particular patient – two examples being a history of substance use disorder and the presence of psychosocial factors. Although certain risk factors associated with drug misuse have been studied, few published studies have used data from a wide variety of multidimensional sources such as occupation, employer types, lifestyle preferences, and other non-pharmacy information to develop models that identify patients at risk for prescription opioid misuse.

Taking a more inclusive, multidimensional view of patient data can help payers go beyond considering the most basic risk factors to truly understand the unique risk profile of their particular claims populations. This enables payers to apply more than just basic intervention strategies, and to customize management to maximize the impact on claims outcomes.

DEVELOPMENT OF A PREDICTIVE MODEL

Healthesystems analyzed 40,000 claims from the last five years to develop a model that could help estimate potential opioid misuse within a given population.2 Misuse in this model was defined as a morphine equivalent dose (MED) higher than the recommended amount.

Because the goal was to identify a more comprehensive set of potential risk factors for opioid misuse, a variety of data types were included in the analysis that may not typically factor into medication management.

THINKING OUTSIDE THE CLAIM

Although some of the factors significantly associated with prescription opioid misuse were expected (e.g., increased days’ supply, early refill patterns), a number of risk factors emerged that have a less-than-obvious association with potential misuse. And importantly, some of these risk factors – such as socioeconomics and lifestyle preferences – are information that isn’t readily visible within a claim.

Once risk factors are identified, we can take it a step further and assign a specific level of risk to each factor to determine which ones are most likely to drive opioid misuse. For the drivers illustrated below, any value higher than 1.0 indicates increased risk for opioid misuse, and any value below 1.0 indicates a reduced risk for opioid misuse.

For example, a patient prescribed opioids by a family physician is at significantly higher risk than the patient prescribed opioids by an orthopedic surgeon. This could be due to a number of factors, including differences in setting (e.g., opioids are typically appropriate for a short-term, post-operative setting), timing (a patient may be under the care of a family physician versus a specialist at different points in the care continuum), or expertise (a primary care physician may not be as well-versed in the management of opioid medications versus a specialist in the field). Despite the reasons, the driver itself is an important insight given that primary care physicians are the top prescribers of pain medications in the United States.3

Drivers of opioid misuse and their varying levels of risk

WHAT MEDICATION MIX REVEALS

While the findings reinforce our need to look at new and more varied aspects of patient data, a predictive model can also enhance our ability to gain insight into a claim’s risk level based on drug mix.

A second analysis uncovered prescription patterns that exist between high-risk and low-risk opioid patients. Claims containing a more complex medication mix were more likely to see an increase in MED. The basic connections that typify these claims are some iteration of the following:

Analytics that consistently demonstrate risk correlation between a specific drug or drug mix can provide much-needed support to guide clinical decision-making in a way that can effectively reduce or even prevent patient risk. As noted earlier, hydrocodone/acetaminophen is one of the most commonly prescribed drugs in workers’ compensation, and the nation’s most popular pain medication. According to the DEA, hydrocodone is also the most abused prescription opioid in the United States. The presence of hydrocodone/acetaminophen as the cornerstone for medication regimens among high-risk opioid claims reinforces the concern that prescribers may underestimate its risks in leading to elevated MED, misuse, and chronic opioid use and dependence.

Analytics that consistently demonstrate risk correlation between a specific drug or drug mix can provide much-needed support to guide clinical decision-making in a way that can effectively reduce or even prevent patient risk.

CURRENT AND FUTURE APPLICATION

The results uncovered reinforce the continued importance of identifying and managing new and potentially overlooked therapy risks, and the need to look more deeply and more comprehensively at patient data. Using pharmacy drug, demographics, employer, and medical claims data, it is feasible to develop predictive models that could assist prescription monitoring programs, payers, and healthcare providers in evaluating patient characteristics associated with elevated risk for prescription opioid misuse.4

The data included in this article originally appeared in the poster Estimating Potential Misuse of Prescription Opioids by Injured Workers in Workers’ Compensation, presented at the American Academy of Pain Medicine (AAPM) 2016 Annual Meeting by Healthesystems Chief Medical Officer, Robert Goldberg, MD, FACOEM.

RATE THIS ARTICLE

SOURCES

1 -Cooper S. Workers Compensation and Prescription Drugs: 2016 Update. NCCI Annual Issues Symposium; May 2016.
2 -Healthesystems. Estimating potential misuse of prescription opioids by injured workers in workers’ compensation. Poster presented at the American Academy of Pain Medicine 2016 Annual Meeting; Palm Springs, CA
3 -Chen JH, Humphreys K, Sha NH, Lembke A. Distribution of opioids by different types of medicare prescribers. JAMA Intern Med. 2016;176(2):259-61.
4 -White AG, Birnbaum HG, Schiller M, et al. Analytic models to identify patients at risk for prescription opioid abuse. Am J Manag Care. 2009;15(12):897-906.
Table of Contents