Cutting Costs and Improving Health
by Robert F. Nease, PhD.,
Chief Scientist, Express Scripts
On the cusp of what will become the biggest change in our nation’s health care system, the debate continues over health care costs and coverage, and who is responsible for ensuring Americans have adequate insurance. There is no doubt that the cost of health care to the nation is staggering, and one of the biggest drivers of this cost is prescription drug coverage.
Although it is a great challenge, this is a manageable cost, specifically when one focuses on the issue of prescription drug adherence, which can lead to reduced waste in medical costs and productivity.
Research has shown that nonadherence can have a profound effect on not only an individual’s health, but on the health care system as a whole, costing up to $100 billion annually. Additionally, 33% to 69% of all medication-related hospital admissions in the United States are due to poor medication adherence, and nonadherence contributes to annual indirect costs exceeding $1.5 billion in lost earning and $50 billion in lost productivity.
Despite the financial cost, few would disagree about the need for patients to take medications as prescribed, and that the rate at which patients are doing this falls short of the ideal.
A November 2009 employer survey from the National Pharmaceutical Council found that medication compliance was a top priority for employers, who are looking to their pharmacy benefit managers (PBM) for solutions. We believe that an advanced application of the behavioral sciences can improve medication adherence. We also believe that because of their role in managing the benefit, their access to patient-level claims data, and their focus on pharmacy, PBMs are uniquely positioned to address this challenge, but only if they realize that nonadherence is primarily a behavioral rather than a clinical challenge.
What is Known and Unknown
Access to detailed longitudinal claims data allows us to determine many observable factors that are associated with adherence— or lack of it. We have found, for example, that each of the following is associated with greater therapy adherence:
- Older age
- Greater income
- Higher disease severity
- Greater number of medications
- Symptomatic disease
- Being partnered
- Partner adherent to maintenance medications
In addition, increasing the cost (or copayment) of medications decreases their use. However, the reverse is true to a far lesser extent: decreasing the cost /copayment for medications has a very modest effect on increasing adherence.
Interestingly, the single most effective intervention proven to improve therapy adherence is using the home delivery channel (i.e., mail-order pharmacy). Studies conducted by Express Scripts suggest that use of this channel independently increases adherence by approximately 800 basis points over the use of retail pharmacies.
A recent report by researchers at UCLA and Kaiser Permanente and published in American Journal of Managed Care found a nearly identical effect: the fraction of patients adherent to medications was 76.9% for retail users and 84.7% for users of mail-order pharmacies.
Just as interesting are the factors that don’t seem to affect adherence. Although most people assume that (at least in general) women are more adherent than men, the data show no such effect.
Another surprise is conscientiousness, one of the “Big Five” personality traits. Conscientiousness is associated with a large number of favorable outcomes: academic performance, job success, engagement in healthy behaviors and avoidance of risky ones, greater quality of life, and enhanced longevity.
Surprisingly, in a large study of personality and pharmacy behavior, we found no association between conscientiousness and adherence. These findings support the hypothesis that for many people, adherence is a matter of habit than of discipline or self control.
Three Main Categories of Nonadherence
The large number of potential causes of nonadherence can be daunting, and as a result we needed an accurate but simple syntax for thinking about nonadherence. In
consultation with our Center for Cost-Effective Consumerism’s Advisory Board, Express Scripts developed a framework for categorizing types of nonadherence.
This framework was later refined based on the results of an analysis of claims data coupled with patient surveys completed in spring 2009. We found that three categories of nonadherence emerged, based on differences in the overall perceived value of the therapy (including cost) and the intentionality of the nonadherence behavior. This framework is illustrated below.
- Sporadic forgetters are those patients who have a positive perceived value of therapy and are not intentional in the adherence behavior. They periodically neglect to take their medications, and as a result are nonadherent.
- Active decliners do not—for a variety of reasons—place a positive value on their therapy. This could be because they believe the medication isn’t effective, are experiencing side effects, don’t like being “prescribed for,” or don’t believe the medication offers sufficient benefit relative to the cost. As a result, they actively choose not to take their medications as prescribed.
- Refill procrastinators do a good job taking their medications as long as they have pills on hand. As their supply dwindles, however, they put off getting a refill. As a result, they may experience a gap in care. This behavior is less intentional than active decliners and does not reflect a negative view of therapy itself.
This simple model provides a powerful language for discussing adherence problems and helping guide the PBM enterprise toward an effective solution.
Key Psychological Principles
Applying the behavioral sciences—including something as simple as altering the words used to communicate with patients— matters when it comes to improving adherence. Express Scripts undertook a large, randomized trial focusing on nonadherent patients in three medication classes: diabetes, cholesterol, and blood pressure. A total of 42,000 members were enrolled in the study and randomized to six different arms (one of which was a control group that received no messaging). The study was designed to determine if strategic messaging utilizing proven psychological principles was more effective in keeping members adherent to therapy.
We found that letters combining authority (letter signed by Express Scripts’ chief medical officer, a physician) and loss aversion (pointing out risks of nonadherence rather than benefits of adherence) out-performed other letters, and significantly increased adherence over the control group. We estimate that to achieve this improvement through the use of financial incentives, copayments would have to be reduced by $10 to $12 per prescription. In other words, words matter.
Cracking the Code
Despite the improvements obtained by messaging based on key psychological principles, PBMs and plan sponsors face three significant challenges in improving adherence. First, multiple causes of non-adherence mean there is no silver bullet solution. Each of the three main categories implies a different approach for addressing nonadherence. Sporadic forgetters may benefit from reminder systems, refill procrastinators from auto-reill programs, and active decliners from financial assistance or a discussion with a clinician. Uniform messaging—no matter how well framed—doesn’t address these patient-level differences.
Second, assessing the effectiveness of interventions to improve adherence is inherently problematic due to something called regression to the mean. Put simply, there is natural variation in adherence over time, and therefore members who are targeted with low adherence at one point in time will tend to improve—even if no intervention is used.
Express Scripts, for example, found that targeting low adherence members and following them but not intervening leads to improvements in medication possession ratio (MPR) of as much as 15%. The implication of this large effect is crucial for plan sponsors to understand: Evaluations of therapy adherence interventions should include a comparable control group. Without such a comparison, it is easy to mistakenly conclude that a worthless intervention is effective. (Note that the randomized controlled trial design used in the therapy adherence messaging study cited above is the gold standard approach; the effects we observed are due to the intervention itself, above and beyond regression to the mean.)
Finally, therapy adherence is generally monitored and measured via claims data. This is problematic because these data are both “laggy” and “lumpy.” By “laggy,” we mean that the data are always a picture of past behavior. Stable estimates of adherence generally require a minimum of three to six months of claims data at the member level.
Thus, today’s measure of adherence does not necessarily paint an accurate picture of today’s adherence behavior. By “lumpy” we mean that the data don’t allow us to determine the type of nonadherence. Put simply, nonadherence due to any of the three main causes look similar from an analysis of the claims data.
Advancing Management of Medication Adherence
Express Scripts is piloting several programs and developing additional capabilities to further improve therapy adherence. One area of tremendous promise is the ability to reliably judge in advance which patients are likely to have adherence problems. Predictive modeling—a mature set of tools and skills—offers the promise of just this ability, and fully leverages the vast amount of data available to PBMs. We believe that such models will serve a crucial role in identifying nonadherence issues before they become bad habits and providing more precise, patient-level targeting. Together, this will mean more effective adherence interventions at a lower cost.
It’s also crucial that we gain a more nuanced and detailed understanding of medication taking behaviors. It is now clear that we must get beyond the claims data if we are to significantly improve our therapy adherence interventions. Exciting new technologies that allow for the measurement of the fine structure of these behaviors in ways that are nonobtrusive to the patient hold out the possibility of gaining this understanding in the very new future.
And these same measurement and monitoring systems, enabled by low-cost digital information management, can serve as a powerful platform for enabling proven principles from the behavioral sciences. Patients, for example, could precommit to sharing their adherence data with their physicians, or have reminders sent to their loved ones. These features offer the potential to more fully leverage the power of social persuasion and social norms to improve adherence behaviors.
It is clear that there is no one-size-fits-all approach to addressing therapy adherence. To combat low adherence, plan sponsors should work closely with their PBM to enact a multifaceted approach to reach the members who are not taking their medications as prescribed whether they are forgetters, procrastinators, or decliners. And these interventions should be subjected to careful scrutiny to ensure that their performance is genuine. But most importantly, we must all understand that, as with much of health care, therapy adherence is primarily about human behavior. As such, the most effective solutions will likely come from those companies that can apply an advanced understanding of how to successfully drive positive behavior change.
Robert F. Nease guides advanced human behavior research at Express Scripts Inc., one of the largest PBM companies in North America. Express Scripts is leading the way toward creating better health and value for patients through Consumerology, the advanced application of the behavioral sciences to health care. More information can be found at www.express-scripts.com and www.consumerology.org.