Managing Opioids, Including Misuse and Addiction, in Patients With Serious Illness in Ambulatory Palliative Care: A Qualitative Study

Abstract

Background:

Pain and opioid management are core ambulatory palliative care skills. Existing literature on how to manage opioid misuse/use disorder excludes patients found in palliative care settings, such as individuals with serious illness or those at the end of life.

Objectives:

We conducted an exploratory study to: (1) Identify the challenges palliative care clinicians face when prescribing opioids in ambulatory settings and (2) explore factors that affect opioid decision-making.

Methods:

We recruited palliative care clinicians who prescribe opioids in ambulatory settings, which included open-ended questions and was conducted online. Results were analyzed qualitatively using a content analysis-based approach.

Results:

Eighty-three palliative care clinicians (mostly MDs/DOs) participated. Challenges faced when prescribing opioids included clinician differences in approach to care (eg, transitioning from another clinician with more permissive opioid prescribing), medication access (eg, inadequate pharmacy supply), resource constraints (eg, access to mental health and addiction expertise), managing problems outside the typical palliative care scope (eg addiction). Participants also discussed factors that influenced their opioid prescribing decisions, such as opioid-related harms and risks that they need to weigh; they also spoke about the necessity of considering other factors like the patient’s environment, disease, treatment, and prognosis.

Conclusion:

This study highlights the challenge of opioid management in patients with serious illness, particularly when misuse or substance use disorder is present, and suggests areas for future research focus. Our next step will be to establish consensus on approaches to opioid prescribing decision-making and policy in seriously ill patients presenting to ambulatory palliative care.

Introduction

Pain management is a common reason for palliative care clinic referral in seriously ill patients¹ and often involves opioid prescribing. In some cases, this can extend to the management of opioid misuse (eg, behaviors such as running out of opioids early, angry/aggressive behavior about opioids, and illicit substance use) and opioid use disorder (OUD, a problematic pattern of opioid use meeting at least 2 of 11 specific criteria including loss of control over use and continued use despite harms leading to significant distress or impairment).² In one national study of ambulatory palliative care clinicians, more than half of respondents reported spending >30 minutes per day managing opioid misuse.³

Recent guidelines for prescribing opioids are based on evidence from the past 10 years which has highlighted the risks of opioid prescribing, especially long-term opioid therapy (LTOT). For example, a recent systematic review suggests that, on average, 25% of patients on LTOT will develop misuse, and 10% will develop OUD.⁴ There is also substantial evidence that LTOT, especially at high doses^5–7 and in combination with benzodiazepines^5,8,9 and gabapentinoids,^10,11 is associated with increased mortality. However, the studies typically exclude patients with serious illness who are often treated in palliative care settings, such as patients with cancer or at the end of life. Guidelines like the Centers for Disease Control and Prevention Guideline for Prescribing Opioids for Chronic Pain¹² explicitly exclude patients receiving “active cancer treatment, palliative care, and end-of-life care.” Although other organizations relevant to patients with serious illness (eg, the American Society of Clinical Oncology¹³) have also released guidelines, their opioid prescribing recommendations are based on the literature for patients without serious illness and therefore may have limitations when applied in palliative care settings.

Therefore, a critical step is to develop an evidence base for palliative care clinicians to guide their care of patients with serious illness who have pain and are at risk for or have opioid misuse/OUD. However, exploratory work is needed prior to this step to identify the most important issues on which to focus research efforts. The objective of this qualitative study was to (1) identify challenges palliative care clinicians face when prescribing opioids in ambulatory settings and (2) explore factors that affect opioid decision-making.

Methods

Study Design

We conducted a national online study of a convenience sample of palliative care clinicians as the first step in a Delphi study, which will develop expert consensus on prescribing opioids for patients in ambulatory palliative care settings.

Participant Recruitment

We e-mailed a survey to members of the American Academy of Hospice and Palliative Medicine (~4575 individuals) and the Hospice and Palliative Nurses Association (~10 197 individuals). We sampled these organizations to recruit palliative care providers in diverse settings around the United States.

Respondents were invited to participate in the survey if they self-reported: (1) prescribing opioids or overseeing opioid prescribing to adults in ambulatory palliative care settings (defined as an outpatient clinic dedicated to providing palliative care, including clinics embedded in specialty care settings) and (2) being at least 18 years old. This study was determined to be exempt by the institutional review board of University of Pittsburgh.

Survey

The survey was developed by the authorship team, which has clinical and research expertise in palliative care, chronic pain including LTOT, and addiction medicine. Notably, of the 6 physician authors, board certifications include palliative care (2), addiction medicine (2), and both (2). It was piloted among the authorship team and other palliative care clinicians practicing in ambulatory settings, and modified iteratively. The survey was administered in September 2018 via Qualtrics 2018 (Provo, Utah). Participants received US$50 for a completed survey.

In addition to demographic information, participants were asked 7 open-ended questions about the challenges faced by opioid prescribers to understand the context and systems in which these clinicians are working, and about opioid prescriber decision-making in order to understand how they make patient-level decisions (see Table 1). Participants were given response boxes with no character limit and were encouraged to “be exhaustive” in their responses.

Table 1.

Participant Survey Questions.

1.	All challenges prescribing opioids in ambulatory palliative care Please identify all challenges you face in prescribing opioids to patients in ambulatory palliative care settings. Please consider situations in which you are continuing, making changes to, or initiating a patient on opioids. Please list all challenges you can think of. Be exhaustive.
2.	Challenges addressing problematic behaviors Patients prescribed opioids in ambulatory settings may exhibit problematic behaviors, also referred to as opioid misuse behaviors, concerning behaviors, or aberrant drug-related behaviors. Such behaviors may include missing appointments, taking opioids for symptoms other than pain, using more opioid medication than prescribed, asking for an increase in opioid dose, aggressive behavior, and alcohol and other substance use. Please identify all challenges you face in addressing problematic behaviors in patients prescribed opioids in ambulatory palliative care settings. Please consider situations in which you are continuing, making changes to, or initiating a patient on opioids. Please list all the challenges you can think of. Be exhaustive.
3.	Opioid-related risks/harms Please describe the opioid-related risks/harms you consider when making decisions about opioid prescribing in ambulatory palliative care settings. Please consider situations in which you are continuing, making changes to, or initiating a patient on opioids. Please list all risks/harms you can think of. Be exhaustive.
4.	Opioid-related benefits Please describe the opioid-related benefits you consider when making decisions about opioid prescribing in ambulatory palliative care settings. Please consider situations in which you are continuing, making changes to, or initiating a patient on opioids. Please list all benefits you can think of. Be exhaustive.
5.	Weighing risks/harms and benefits In the prior question, you identified opioid-related risks/harms and benefits you consider when making decisions about opioid prescribing in ambulatory palliative care settings. Please describe how you go about weighing these risks/harms and benefits. Please consider situations in which you are continuing, making changes to, or initiating a patient on opioids. Please list as many approaches you can think of. Be exhaustive
6.	Factors that lead to a tailored approach Think about the approaches you described in the prior question. Consider that these approaches might be tailored to an individual case based on a variety of factors, including patient, provider, caregiver, system, and other factors. Please describe the factors that might lead you to tailor your approach to a particular patient in an ambulatory palliative care setting. Please consider situations in which you are continuing, making changes to, or initiating a patient on opioids. Please list all the factors you can think of. Be exhaustive.
7.	Prognosis impact on approach Please describe how, if at all, patient prognosis impacts your approach to weighing opioid-related risk/harms and benefits when making decisions about opioid prescribing in ambulatory palliative care settings

Sample Size

The purpose of our study was 2-fold: (1) to conduct exploratory work needed to identify the most important issues on which to focus research efforts, and (2) to serve as the initial brainstorming round of a Delphi study. The purpose of this Delphi study will be to establish consensus among experienced clinicians on how to address the issues identified in the present study: this field’s most pressing clinical and policy challenges. We chose our sample size with both purposes in mind. Delphi study sample sizes range widely, but between 15 and 20 participants is considered a minimum number that allows for diverse viewpoints and also dropout between rounds.^14,15 A somewhat larger Delphi panel (eg, 40) allows for dropout between rounds.¹⁶ This is larger than sample sizes often needed to reach thematic saturation in qualitative analyses. Therefore, 40 was our target sample size.¹⁷

Analysis

We report descriptive statistics using frequencies or medians and interquartile ranges. Four coauthors with diverse backgrounds (2 palliative care physicians: J.S.M., J.C.; 2 PhD-trained qualitative methodologists: S.R.Y., K.F.G.) reviewed the first 25 transcripts to identify an initial coding approach, which fits guidelines of having a group up to 5 people.¹⁸ One methodologist (S.R.Y.) and one physician (J.S.M.) then independently coded responses using a content analysis-based approach.¹⁹ We chose this approach because we anticipated that saturation for deeper meaning would be difficult in a one-sided survey in which follow-up probing was not possible. After initial coding, we adjudicated discrepancies and clarified code definitions in order to ensure uniform application of the codebook.^18,20 During this process, codes were grouped based on the questions participants were responding to: (1) opioid prescriber challenges (questions 1 and 2), and (2) opioid prescriber decision-making (questions 3, 5, 6, and 7). Because of the nature of the questions, there was overlap in content between opioid prescriber challenges and opioid prescriber decision-making. To frame the response in the appropriate context, we coded the response based on the question the participant was answering. Question 4 (opioid-related benefits) was analyzed but given the straightforward nature of the question and answers received is not included in the results below. The remaining transcripts were divided between the 2 coders who applied the consensus codebook.^18,21 NVivo version 12 was used to manage the analysis (QSR International Pty, version 12; 2018).

In the results, we only expand on categories and subcategories that were commonly endorsed and are important for furthering the broader research in this area. However, in Supplemental Tables, we present the comprehensive list of categories and subcategories along with their definitions and exemplary quotes. This complete list of categories and subcategories may have future importance to palliative care clinicians, researchers, and policymakers.

Results

Eighty-three palliative care clinicians completed surveys within the first 12 hours, more than double our target sample size, at which point data collection was stopped. They were majority white, female, and MDs or DOs (Table 2). Slightly more than half were fellowship-trained in Palliative Medicine. Both academic and community settings were well represented (43% and 31%, respectively).

Table 2.

Participant Characteristics.^a

Characteristic	Frequency, % or Median (IQR)
Age	42 (36–57)
Race
White	67 (80.7%)
Black or African American	4 (4.8%)
Asian	12 (14.5%)
Gender
Female	52 (62.7%)
Male	31 (37.3%)
Highest clinical degree
MD	66 (79.5%)
DO	8 (9.6%)
NP	4 (4.8%)
Other	2 (2.4%)
Residency training
Internal medicine	43 (51.8%)
Family medicine	19 (22.9%)
Psychiatry	1 (1.2%)
Pediatrics	8 (9.6%)
Other	7 (8.4%)
Not applicable, not a physician	6 (7.2%)
Fellowship training
Palliative medicine	47 (56.6%)
Internal medicine subspecialty	11 (13.3%)
Pain medicine	1 (1.2%)
Other	8 (9.6%)
Not applicable, not a physician	6 (7.2%)
Not applicable, no fellowship training	17 (20.5%)
Completed last postgraduate degree (residency, fellowship, NP degree)
Less than 5 years ago	29 (34.9%)
More than 5 but less than 10 years ago	20 (24.1%)
More than 10 but less than 15 years ago	10 (12.0%)
15 years or more ago	24 (28.9%)
Current clinical setting (select all that apply, N - 101)
Academic	43 (42.6%)
Community	31 (30.7%)
VA	10 (9.9%)
Other	17 (16.8%)b
Years in current clinical setting	5 (2, 9.3)
Composition of practice	(mean percentages)
Cancer	62
Heart failure	8
Chronic obstructive pulmonary disease	8
Neurologic conditions	7
Dementia	7
End-stage renal disease	3
Other	6
Waivered to prescribe buprenorphine	9, 10.8%

Abbreviation: VA, Veterans Health Administration Hospitals.

^aN = 83.

^b“Other” includes group practices, multispecialty practices, long-term care, home-based care, and hospice.

Opioid Prescriber Challenges Described by Participants

Care approach differences.

Participants described the difficulty in managing the expectations of patients who transition from another care team with different opioid prescribing practices. For example, one participant reported, “Cancer survivors who are free of disease whose oncologist has been prescribing high dose opioids for many years, only to decide that this is not a good idea, then refer those patients to PC [palliative care] clinic for ‘weaning.’…. patients not on board once they come to clinic.” Participants also noted that referring clinicians expect the palliative care clinic to manage opioid misuse: “Oncologists do not want to prescribe for patients with history of misuse or abuse and so expect our team to manage this, despite [patients] having cancer or treatment related pain.” These divergent clinician views related to opioid prescribing can lead to patients getting conflicting care: “sometimes when I have a long talk with a patient about their use, my concerns, alternatives, etc. they just go next door to the Oncologist who then gives them as many [P]ercocets as they want.”

Medication access.

This included access to a pharmacy with adequate supply, affordability and cost, and insurance coverage issues such as requiring administrative hurdles to prescribing (eg, prior authorizations). Participants also described instances where the dispensing pharmacy did not honor prescriber decisions, which can lead to bias: “Outpatient pharmacists sometimes refuse to fill prescriptions for opioids without calling to talk to me about it. This seems to be disproportionately done on a racial basis.”

Resource constraints.

Resource constraints included access to pain, addiction, and mental health specialists, lack of clinician and staff time to address pain and opioids, and limited palliative care clinician expertise in managing misuse and addiction.

Systems factors.

Participants identified systems factors at the level of the participant’s organization, as well as broader regulatory policies.

Organizational environment.

Respondents cited many examples of lack of organizational support, particularly for managing patients with substance use disorder. One participant stated, “Hospital [administration] and [Department] chair not addressing issues of substance abuse in cancer population. (Refusing to hire addiction medicine specialists).”

Regulatory policy.

Participants described challenging regulations outside of their institution, often at the state level, such as obtaining informed consent and checking the prescription monitoring database with every fill.

Managing patients outside the typical palliative care scope.

Participants reflected on how some of the challenges they face may be outside of the typical palliative care scope. This can include being referred patients with chronic pain who do not have a serious, life-limiting illness or “specialty palliative care needs,” a concept which participants noted is ill-defined in the field. As one participant reflected, “we in the field have big, generous, nonsuspecting hearts and are prone to being duped.”

Concurrent palliative care needs and substance use.

One aspect of managing patients outside of the typical palliative care scope is managing patients who have palliative care needs and concurrent substance use. One participant described the challenge of “Walking the fine line between safe prescribing and being compassionate while managing cancer patients with active or [history] of substance abuse.”

Participants’ Descriptions of Opioid Prescriber Decision-Making

Weighing opioid-related harms and risks.

Harms included respiratory depression and other side effects. Risks included antecedent or baseline factors, many of which are basic information about the patient’s history: current or past substance use, psychiatric illness, opioid misuse, criminal justice involvement, housing insecurity, health literacy, patient understanding of opioid risks, and other medications that have been tried. One participant explained how they incorporate these potential risks into their prescribing decisions: “If a patient is found to be high risk, we tend to give them fewer pills and see them more frequently with more frequent urine drug screen [UDS] and prescription drug monitoring reports. If UDS is positive for cocaine, heroin[,] etc[.], we stop prescribing unless they are dying.”

Patient environment.

Participants reported considering a number of factors about a patient’s environment when deciding to prescribe opioids. This could include the patient’s social support, negative family influence (eg, “Worry about family members with Substance Use Disorder taking medication”), and the potential danger posed by opioids in the home (eg, “Keeping meds safe from minors in the home” and “Creating target for crime/burglary”). One participant explained how factors such as an “involved and supportive family [that] can help monitor for side effects and improve early communication if problems arise” can influence their decisions, and result in an “increased comfort level with changing/escalating/rotating regimen.” Another participant described the opposite scenario: “In a patient with less support at home…I would be more inclined to be more conservative with dose titration.”

Factors related to the disease or its treatment.

Participants noted other factors that affected their decision-making: the patient’s diagnosis (eg, “cancer type”), performance status, pain etiology (eg, “cancer vs chronic pain”), and treatment type, including “whether the patient has completed other treatments aimed at reducing pain (eg, radiation).” Participants also mentioned a patient’s “probable disease course” including whether they are “disease-free.” For example, one participant stated “When patient has curative intent of therapy, I am more conscious of the need to limit frequent escalations of medication due to future expectations to wean and/or stop opiate use in the future.”

Prognosis.

Some participants discussed being less concerned about opioid-related risks and monitoring close to the end of life. Though some participants reported using tools to assess risk on everyone, most participants reported that they prescribe opioids more liberally in patients with a short prognosis: “If the patient is likely to die, we are more lenient. If the patient is dying, we are the most lenient.” However, the length of time participants used to describe a short prognosis varied from 3 months to anything less than a year. One participant noted, “[w]hile we don’t have a clear, arbitrary cutoff, we are certainly more willing to prescribe for patients with shorter prognoses than longer.”

Managing opioid misuse/OUD at the end of life was not typically considered a priority, although there were exceptions, such as when a patient preferred to limit their own opioid use. They also mentioned instances of monitoring patients more closely using home visits or considering stopping opioids if they strongly suspected diversion.

Other participants did not think prognosis is an important factor in opioid-related decision-making. As one participant put it, “Risks/harms and benefits don’t go away or lessen if patient has a poor prognosis.”

Discussion

This national study of 83 palliative care clinicians highlights the challenge of opioid management in patients with serious illness, particularly when misuse or substance use disorder are present, and suggests areas for future research focus. Our participants described a system in which referring clinicians manage the patient’s serious illness, but defer pain management and opioid prescribing to palliative care due to wariness about managing opioids in the current climate. In addition, participants reported that referring clinicians assume palliative clinicians have expertise in managing opioid misuse and substance use disorders, despite studies suggesting significant gaps in training and knowledge in these areas.³ In fact, some participants raised the question of whether treating opioid misuse and substance use disorders in patients with serious illness should even be within palliative care’s scope. These palliative care clinicians reported that referring to appropriate specialty (eg, mental health, substance use disorder, and chronic pain clinics) or primary care clinics to manage opioid misuse, OUD, and related comorbidities is severely limited by resource constraints. This problem is compounded by these clinics often being unwilling or unable to manage patients with concomitant serious medical illness.

Whether opioid misuse and addiction is—or should be—within the scope of palliative care practice, palliative care clinicians are actively confronted by these issues.³ Our findings highlighted some key management considerations. For example, most participants reported being more “liberal” with opioid prescribing in patients with serious illness in ambulatory palliative care settings, although some do consider it important to address opioid misuse and substance use disorder even at the very end of life. Participants also described patient-level factors that impact their decision-making, including diagnosis (eg, cancer type) and likely disease trajectory (eg, potential for long-term survival). However, the limited evidence on the best way to incorporate these patient-level factors into decision-making is noteworthy.²²

Participants also discussed factors that are completely out of the patient’s control, such as their living situation or the presence of a caregiver interfering with their care or even diverting their medications. This is a consideration unique to opioid pain management, as there are few other medications potentially diverted by family and other caregivers. Because caregivers are often very involved in managing medication, this may lead to suboptimal pain care in patients who do not have flexibility (eg, alternative caregivers or living situations) due to poverty or other life circumstances. This scenario deserves future investigation.

Our findings also draw attention to how responses to the opioid epidemic intersect with the needs of seriously ill patients prescribed opioids. As national attitudes toward opioids have changed, a variety of measures meant to slow the epidemic have also impacted pain management practices. As noted by our participants, this is reflected in state-level regulatory scrutiny and lack of access to opioid medications.²³ Though these limitations were originally intended for patients with chronic “non-malignant” pain, they have reached patients with serious illness in palliative care settings. Whether these limitations are appropriate remains to be seen; thus, future research should seek to understand the systems-level approaches that support optimal pain management in patients with serious illness in ambulatory palliative care, while also mitigating opioid risks and ensuring the treatment of OUDs. Taken together, we hope that this study’s findings provide a starting point to develop consensus on management strategies.

These findings should be considered in light of the following limitations. First, we included only palliative care clinicians. While it is instructive to understand the challenges palliative care clinicians face and what factors influence their decision-making, most palliative care clinicians are not experts in chronic pain and opioid misuse or use disorder. Future studies, especially those that seek to develop consensus on optimal management strategies, should be more inclusive of specialties such as addiction medicine. Specialties that focus on the treatment of specific serious illnesses, for example, oncology, will also be important to include. In addition, patients with serious illness may receive opioid management in other settings, including primary care and pain clinics, where our findings may not be applicable. Also, nurses who are involved in opioid prescribing (eg, nurse practitioners) and who make up a significant proportion of the ambulatory palliative care workforce represent a small proportion of respondents, despite our recruitment efforts. Further, our respondents are not representative of palliative care providers nationally, but are a convenience sample. Finally, this study’s approach was limited by its online format, as there was no opportunity to deeply explore responses with probing.

In sum, this study is an important step toward defining the challenges of opioid management in ambulatory palliative care settings. Future work should endeavor to improve prescribing processes of care and develop guidance and tools to support frontline clinicians and policymakers. As mentioned above, this study was designed so that it could be the first round of a Delphi study, the goal of which would be to use these findings as a starting point to establish consensus on optimal approaches to opioid prescribing decision-making and policy in seriously ill patients presenting to ambulatory palliative care settings.