Promoting Patient-Centeredness in Opioid Deprescribing: a Blueprint for De-implementation Science

Abstract

A downward trend in opioid prescribing between 2011 and 2018 has brought per-capita opioid prescriptions below the levels of 2006, the earliest year for which the Centers for Disease Control and Prevention has published data. That trend has affected roughly ten million patients who previously received long-term opioid therapy. Any effort to reduce or replace a prior health practice is termed de-implementation. We suggest that the evaluation of opioid prescribing de-implementation has been misdirected, within US policy and health research, resulting in detrimental impacts on patients, their families and clinicians. Policymakers and implementation scientists can address these deficiencies in how we study and how we perform opioid de-implementation by applying an implementation science framework: the Consolidated Framework for Implementation Research. The Consolidated Framework lays out relevant domains of activity (internal, external, etc.) that influence implementation processes and outcomes. It can deepen our understanding of how policies are chosen, communicated, and carried out. Policymakers and researchers who embrace this framework will need a better approach to measuring success and failure in health care where both pain and opioids are concerned. This would involve shifting from a reductive focus on opioid prescription counts toward measures that are more effective, holistic, and patient-centered.

A downward trend in opioid prescribing has emerged in the wake of a crisis of overdose and addiction related to opioids, coupled with a recognition that the benefits of opioids had been oversold.¹ According to 2019 data from the US Centers for Disease Control and Prevention (CDC), opioid prescribing has declined 37% since its peak in 2011, while prescriptions exceeding 90 morphine milligram equivalents fell by 67% since 2008.² This decline accelerated after a 2016 Guideline was issued.³ Such efforts, however, can have unintended consequences.⁴ By 2019, there was widespread recognition that reductions in prescribing had not been implemented in ways that consistently protected patients.^5–9 Patients, media, government agencies, and professional literature acknowledged instances of worsening pain, loss of access to care, and death by suicide,^7,10–13 even as others described successes in patient-centered voluntary dose reduction¹⁴ and post-operative pain management.¹⁵

By 2019, the CDC avowed that its guidance had been misapplied.^16,17 Just as opioid prescribing once had rocketed ahead of the supporting science, opioid deprescribing was often mandated and carried out in ways that lacked evidentiary support.^6,7,18,19 But this failure raises urgent questions for agencies seeking to implement changes to care and for researchers who partner with them: how can well-intended changes to care sometimes transpire in ways that are unsafe or harmful? Studies that measure the rise and fall of prescriptions rarely portray how such opioid prescribing changes are carried out.²⁰ So how should these changes to care be measured, particularly if patient harm may be taking place amidst efforts to change care that are intended to be beneficial?²¹

Implementation science provides a framework to explain how organizational efforts to change care may play out poorly in practice and can improve efforts to measure such change.^22–24 In this perspective, we describe how a well-regarded framework for understanding implementation of health system change can guide needed changes and ensure that research can assess their success or failure for the 10 million Americans who receive opioids long term.²⁵

CONSIDERATIONS TO GUIDE FUTURE IMPLEMENTATION SCIENCE ON PRESCRIPTION OPIOID REDUCTION

Research on changing opioid prescriptions has emphasized number of pills prescribed,²⁶ or the number of long- versus short-term prescriptions,²⁷ with weaker focus on outcomes for patients, or on how health care organizations reduce or eliminate opioid prescribing for long-term recipients. De-implementation has been defined as reducing or stopping services or practices that are ineffective, unproven, harmful, overused, or inappropriate.⁴ In relation to prescription opioids, de-implementation can mean dose reductions, stoppage, or a decision not to initiate. In proposing how better to study this process, one stipulation is offered. The interest in reducing opioid reliance does not represent, for us, a commitment to eliminating opioids from pain care.²⁸ Rather, if de-implementation occurs, it must be undertaken with care,^4,29 so as to alleviate adverse outcomes^7,12,13,30 and avoid exacerbating health care inequities.³¹ The following section will show how the Consolidated Framework for Implementation Research (CFIR) can help us carry out, evaluate, and study those changes in a way that includes patients.³²

CONSOLIDATED FRAMEWORK FOR IMPLEMENTATION RESEARCH

Implementation researchers ask how and why efforts to change health care delivery succeed or fail. It is not enough to ask whether prescriptions have declined, but how reductions were made to happen and what the impact was on patients with pain. The CFIR groups relevant constructs into five domains.³² Each domain can shed light on shortfalls in current opioid reduction efforts and point to contextual factors that influence whether de-implementation is done well or poorly. For these reasons, metrics serve as the focal point for recommendations to improve research in the future.

Characteristics of Individuals

This domain is linked to all others, as the prescriber and collaborating clinical personnel are central actors in any deprescribing process. This domain includes professionals’ knowledge and competence with pain, opioids, and related topics, along with the condition of their therapeutic relationship. It includes their comfort with difficult interactions. Clinicians’ longstanding distrust of patients with pain or opioid receipt,^33,34 and their discomfort with uncertainty,³⁵ may pose challenges. One study found primary care providers often are not motivated to treat chronic pain or opioid use disorder.³⁶ Finally, the prescriber’s ability to handle external pressures matters. Such pressures may include fear of losing a license, being investigated, being physically attacked, or being censured. Training initiatives address knowledge,³⁷ but rarely address provider capacity to handle external pressures or to achieve clinical engagement and boundary clarity.³⁸ Future de-implementation efforts should evaluate provider characteristics, such as knowledge and competence related to taper and pain care, confidence and ability to navigate challenging professional pressures, and how these factors influence de-implementation and patient outcomes.

Characteristics of the Intervention

To study the effects of local, regional, or national prescribing policies, researchers must document specific steps taken to alter care. They must characterize the flexibility of the policy (or alternatively, its inflexibility). What letters, orders, policies, resources, supports, or other actions were used to change pain care and opioid prescribing? What did clinicians perceive to be the consequences of deviating from new policies? Health system payers and administrators often are the ones who decide whether de-implementation of some practices will be coupled to implementation of others,³⁹ such as new pain care modalities.^40,41 They decide whether prescription reduction policies are unilaterally mandated, or permit individualization. Given the lack of evidence on across-the-board opioid reduction outcomes, we favor strategies that attend to the unique needs of the individual patient. Whether individualization is permitted, however, is rarely reported.^20,42–45 Evaluation of de-implementation efforts should ask whether proposed alternatives to opioids are appropriate for the patient, provider, and setting.²⁹ Such research should include record review, surveys, interviews, and organizational assessments to capture these matters.

Inner Setting

Inner setting includes organizational culture and actions taken by leaders including commitment and alignment of effort.⁴⁶ Their actions may include offers of tangible support, or the use of threats or reward to accelerate changes, or to pre-empt expressions of concern from staff. Care transformation efforts falter if perceptions of professional risk constrain communication about risk to patients.⁴⁷ The CFIR defines “learning climate” to include psychological safety: the degree to which “leaders express their fallibility and need for team members’ assistance and input.” Such issues echo across industries and they are measurable.^48,49

Process

The implementation process includes multiple organizational factors, such as implementation plans, engagement of the team, resource allocation, results measurement across stakeholders, and how the process is revised in real time. Contemporary reports on opioid deprescribing often lack descriptions of process of change,^20,21,43 or they reference a single element (e.g., a clinic policy)⁵⁰ with prescription volume and dose treated as a proxy for both the intervention and the outcome. Such “pill dynamic” analyses typically fail to include measures of process or patient-centered outcomes, and they fail to measure the experience of the prescriber.^20,43,51 As a result, they tend to reinforce a reductive focus on opioid dose reduction, alone.

Outer Setting

Outer setting is key to understanding why opioid prescribing reductions unfolded in problematic ways in the USA.^6,7,13,52 Outer setting has several components, but we focus on two.

The first domain involves the extent to which the implementation (or de-implementation) program reflects the needs of the patient group and the barriers to meeting such needs. It is axiomatic that patient needs should be the first consideration in care delivery, and yet programmatic efforts to address pain may falter in this regard. Patients with severe pain often have multiple serious comorbidities⁵³ and logistic barriers that may stymie receipt of care.⁵⁴ The complexity of the patient and their health care needs is insufficiently addressed in national opioid prescribing guidance.^55–57 When opioid reduction is considered, one published tapering protocol stresses the need for patient-centeredness, including in the outer setting: “the focus should never be solely on opioid reduction. The biopsychosocial model of pain treatment should be applied to facilitate not only goals for pain and opioid reduction but improved function and quality of life.”⁵⁸

The second aspect of the outer setting involves external policies. For instance, national quality metrics have caused problematic forms of de-implementation rife with unintended consequences. External policies, including metrics, reflect public concern, which has taken count of the role played by prescriptions in the overdose crisis.^59,60 That concern is embodied by high-stakes litigation,⁶¹ state and federal laws, novel regulations,⁶² media coverage,⁶³ research, and payer policies. One quality metric treats the number of patients at high dose (initially, 120 MME, later revised downward to 90 MME) as reflecting poor care. That metric was proposed by the Pharmacy Quality Alliance,⁶⁴ then embraced by the National Committee for Quality Assurance⁶⁵ and others.⁶⁶ In the CDC’s Guideline, the 90 MME threshold was couched as cautionary (“should avoid increasing dosage to ≥90 MME/day or carefully justify a decision to titrate dosage to ≥90 MME/day”), and the CDC did not require or insist on dose reduction for patients above such doses.⁵⁷ While invoking the CDC’s authority, the Department of Health and Human Services Office of the Inspector General simplified 90 MME as a dose “to avoid” in a report declaring its partnership with law enforcement. That threshold is now used to profile federal agency performance,⁶⁷ to allocate bonus payments under the Medicare 5-Star Program,⁶⁸ to initiate criminal investigations,⁶⁹ and for pharmacy benefits decisions. The original distinction between doses forced down, versus doses not escalated, has no presence in these metrics used by law enforcement, quality arbiters, or payers. This situation induces fear among clinicians. In interviews conducted by our team, one clinician stated: “Government involvement in this issue really does not help in terms of regulation, it makes people paranoid about treating patients.”⁷⁰

Contemporary scholarship on policy helps to explain why implementation of the CDC’s Guideline has often contradicted both the language and the spirit of the Guideline itself.^7,71 First, policy scholars note that the aspiration for completely rational policies is almost never upheld in real-world responses to complex problems (this is termed “bounded rationality”⁷¹). Decision-makers lack the time to absorb the problems they are called upon to solve, or power to fully solve them. Thus, they are drawn to reductive solutions, targeting the most accessible point of leverage, based on which advocates they listen to. Second, policy is not shaped by one party, but by many at once. We illustrate the diversity of opioid prescription policy vectors in Figure 1, broken crudely into categories of “Governmental,” “Guidelines and Metrics,” “Providers and Payers,” and “Framing Voices,” even though there is overlap among these compartments. Very few studies formally assess how these parties interact and the resultant effects on care.⁷²

Fig. 1.

The figure offers selected examples of actors and agencies that, collectively, influence opioid prescribing in the USA, grouped into the following domains: “Governmental,” “Guidelines and Metrics,” “Framing Voices,” and “Providers and Payers.” The four domains are offered with illustrative intent and are not necessarily exhaustive or mutually exclusive. For example, a government agency may produce a guideline and also influence how the guideline is understood through collaboration with a journalist (i.e., a Framing Voice). CMS, The Centers for Medicare & Medicaid Services; CDC, Centers for Disease Control and Prevention; DEA, Drug Enforcement Agency; VA, Department of Veteran Affair’s; DoD, Department of Defense; NCQA, National Committee for Quality Assurance.

A WAY FORWARD

The CFIR domains should be considered by leaders and researchers seeking to advance pain care. The domains expose the complexity of prescription opioid de-implementation and its impact on patients. Better metrics could help capture that impact. Table 1 offers examples. At the health system level, one measure deserving priority is the count of patients lost from clinics or health systems. Patients often feel forced to leave clinics when their doctors announce that prescription opioids will be reduced or stopped.⁷³ One clinic reported losing 36% of its patients.⁷³ Other proposed measures, including mortality, are offered in Table 1. They include patient-reported autonomy in taper, readiness to taper, satisfaction with behavioral treatments, and functional outcomes.⁵⁸ While not exhaustive, that list can guide health systems and research partners to develop programs focused on best serving patients with pain, related vulnerabilities, and opioid receipt, and better stewardship for their care.

Table 1.

Enhanced Metrics for Success in Assessing Prescription Opioid de-implementation

Metric	Justification	Anticipated challenges
Systems-level metrics
Mortality after opioid dose change1	Mortality increase after opioid stoppage seen in some reports	Specific causal attributions subject to debate
Hospitalization within 3 months of opioid dose change1	Emergent hospital use seen in some studies	Disentangling contributors to hospitalization may be difficult
Number of patients accessing alternative pain care1	De-implementation of opioids requires a plausible pain care alternative	Denominator of persons needing care may be difficult to calculate
Number of patients accessing mental health services when there is a coexisting mental health diagnosis1	Mental symptoms influence pain experience	Not all mental health conditions require professional treatment
Number of patients leaving health system after opioid stoppage1	Termination of care relationships noted in some reports	Re-engagement after termination of care has not been studied
Patient-level metrics
Appropriateness of dose3	The CDC Guideline calls for documenting calibration of risk and benefit based on patient function	Divergent views on opioids’ utility in chronic pain
Pain expectations2	Patient expectations influence care experience and retention	Patient expectations vary over time
Opioid dose expectations	Expectations regarding opioid prescribing are distinct from expectations about pain	Patient expectations vary over time
Taper autonomy, as perceived by patient	Perceptions of losing control in care may be traumatic	Not all patients require or expect full autonomy in medication choice
Satisfaction with behavioral treatments	Patient engagement and retention depend on satisfaction with treatment	Satisfaction may not indicate quality of the treatment
Use of other illicit drugs or alcohol in context of opioid prescription change	Transition to non-prescribed substance may confer additional risk	Self-report and laboratory indicators have varying levels of accuracy over time

CONCLUSIONS

When a multi-faceted, complex health issue becomes a public health crisis, the desire to “solve” or “mitigate” takes hold with a momentum of its own. A crisis deserves no less. However, nationally adopted quality metrics have convinced some patients with pain that their survival and functioning are no longer concerns for the systems in which they receive care. This outcome is unacceptable.

Health systems must measure outcomes of their de-implementation efforts to ensure that their actions are advancing patient health. Implementation science points toward a reappraisal of how our health system responses to the opioid crisis can become more effective, holistic, and patient-centered.

Compliance with Ethical Standards

Conflict of Interest

Dr. Kertesz affirms past personal ownership of stock in Merck and Abbott pharmaceutical companies, sold in December of 2017, and not exceeding 5% of his assets at the time. Dr. Kertesz holds stock in CVS Caremark, Thermo Fisher, and Zimmer Biomet, not exceeding 5% of his assets. Dr. Kertesz reports his spouse holds equity in Merck, Abbot, Thermo Fisher, and Johnson and Johnson, in her private assets, not exceeding 10% of her assets. Dr. Kertesz receives income from UpToDate, Inc. Dr. Darnall is the recipient of a research award from the Patient-Centered Outcomes Research Institute® (#1610-3700), author royalties from multiple books, and compensation as chief scientific advisor at AppliedVR. Dr. Varley receives part-time income from Heart Rhythm Clinical and Research Solutions.

Disclaimer

The views expressed in this article are those of the authors alone and do not represent positions or views of the United States Department of Veterans Affairs. All statements in this report are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute (PCORI), its Board of Governors, or Methodology Committee.