Healthcare provider knowledge, beliefs, and attitudes regarding opioids for chronic non-cancer pain in North America prior to the emergence of COVID-19: A systematic review of qualitative research

ABSTRACT Background Balance between benefits and harms of using opioids for the management of chronic noncancer pain (CNCP) must be carefully considered on a case-by-case basis. There is no one-size-fits-all approach that can be executed by prescribers and clinicians when considering this therapy. Aim The aim of this study was to identify barriers and facilitators for prescribing opioids for CNCP through a systematic review of qualitative literature. Methods Six databases were searched from inception to June 2019 for qualitative studies reporting on provider knowledge, attitudes, beliefs, or practices pertaining to prescribing opioids for CNCP in North America. Data were extracted, risk of bias was rated, and confidence in evidence was graded. Results Twenty-seven studies reporting data from 599 health care providers were included. Ten themes emerged that influenced clinical decision making when prescribing opioids. Providers were more comfortable to prescribe opioids when (1) patients were actively engaged in pain self-management, (2) clear institutional prescribing policies were present and prescription drug monitoring programs were used, (3) long-standing relationships and strong therapeutic alliance were present, and (4) interprofessional supports were available. Factors that reduced likelihood of prescribing opioids included (1) uncertainty toward subjectivity of pain and efficacy of opioids, (2) concern for the patient (e.g., adverse effects) and community (i.e., diversion), (3) previous negative experiences (e.g., receiving threats), (4) difficulty enacting guidelines, and (5) organizational barriers (e.g., insufficient appointment duration and lengthy documentation). Conclusions Understanding barriers and facilitators that influence opioid-prescribing practices offers insight into modifiable targets for interventions that can support providers in delivering care consistent with practice guidelines.


Introduction
Chronic noncancer pain (CNCP) is considered one of the most prevalent, debilitating, and complex medical conditions to manage. 1 Though estimates vary depending on survey methodology, nationally representative data from North America and Europe indicate that 20% to 30% of adults experience CNCP. 2,3 Urgency of patient needs, aggressive marketing to promote widespread use, demonstrated effectiveness of opioids for acute pain, 4 and limited access to alternative therapeutic modalities have perpetuated the use of opioids for the treatment of CNCP. 5 In the United States, between 1997 and 2011, there was a fourfold increase in the sale of prescription opioids, 6 fivefold increase in drug treatment admissions for prescription opioids (from ~20,000 to ~120,000), 7 twofold increase in emergency department visits related to pharmaceutical opioids, 8 and fourfold increase in opioid-related overdose. 9 Though commonly prescribed, evidence for the benefits of opioids for CNCP is modest (risk difference of achieving the minimally important difference in pain relief versus placebo is 12%), 10 and potential risks range from nausea and vomiting to addiction, overdose, and death. 5 Guidelines for prescribing opioids for CNCP are adopting increasingly conservative recommendations in order to balance potential risks against uncertain evidence for benefit. The most recent guidelines published in the United States 11 and Canada 12 have been viewed by some as overly conservative and were criticized for being introduced into health care systems that were ill-equipped to provide effective alternatives, such as nonpharmacological interventions for people who live with CNCP. [13][14][15][16] For example, these guidelines recommend against prescribing for people with mental health issues or young people, which could create inequalities in the receipt of quality care among these populations. 15 Three out of four individuals who use heroin report that their misuse of narcotics began with prescription opioids. 17 The opioid-related mortality rate in Canada in 2016 was 7.9 per 100,000 population, 18 with 1 in every 91 deaths being attributed to opioids. In 2019 the rate of opioid-related mortality per 100,000 increased from 7.9 to 11.9. Providers experience an understandable reluctance to prescribe opioids to patients with CNCP given the impact of substance misuse; however, strict opioid prescribing due to concerns over substance misuse can also lead to patient harms (e.g., increased pain, risk of depression, and decreased quality of life). 15 Patients whose pain is not managed because of underprescribing are also more likely to turn to drug diversion or obtaining illicit opioids in order to relieve pain, placing them at an increased risk of potential overdose. 15 Health care providers express understandable uncertainty regarding whether, when, and how to prescribe opioids for CNCP given the trade-off between risks and benefits and conflicting evidence on the impact of patients from prescribing or withholding opioids.
It is worth noting that during the first year of the pandemic, there was a 95% increase in apparent opioid toxicity deaths (April 2020-March 2021, 7224 deaths) compared to the year before (April 2019-March 2020, 3711 deaths). 19 Several factors may have contributed to a worsening of the overdose crisis over the course of the pandemic, including the increasingly toxic drug supply, increased feelings of isolation, stress and anxiety, accessibility of services for people living with pain, delays in medical care for people living with pain, and reduced access to harm services for people who use drugs. Though it is evident that COVID-19 had an impact on opioid-related deaths and a potential impact on prescribing practices, it is important to synthesize evidence pertaining to provider attitudes toward prescribing opioids for CNCP before the pandemic as we transition into the endemic phase of the COVID-19 crisis. Such knowledge can serve as a starting point for understanding the barriers and facilitators of opioid prescribing independent of the significant impact of a global pandemic.
It is important to gain a rich understanding of the barriers and facilitators associated with prescribing opioids for CNCP given that such knowledge is vital for (1) understanding behavior, (2) developing behavioral interventions, and (3) implementing sustainable behavior change in a manner that improves the alignment between evidence and practice. One qualitative synthesis has been published that explored provider experiences of prescribing opioids for CNCP but did not focus on gaining a rich understanding of barriers or facilitators that could be used to develop and enact interventions to modify prescribing practices. 20 The purpose of this review was to address this gap in the literature by constructing an understanding of barriers and facilitators for prescribing opioids for CNCP through a systematic review of qualitative literature reporting on opioid prescribing for CNCP. Further, it is important to have a synthesis of literature published before the COVID-19 pandemic given that such knowledge is tantamount to understanding changes that may have occurred as a result of the global pandemic.

Search Strategy
This review is reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis 21 and Enhancing Transparency in Reporting of Synthesis of Qualitative Research. 22 The protocol for this review was preregistered (PROSPERO #CRD42018091640) and published. 23 Our search strategy was developed by an information specialist in consultation with content experts and subject to peer review by an independent information specialist. A systematic search of CINAHL, Embase, MEDLINE, PsycINFO, and Cochrane CENTRAL was performed from inception to June 3, 2019. We searched the Cochrane Library databases and the Joanna Briggs Institute for relevant systematic reviews and PROSPERO for relevant registered protocols. Manual searches of the reference lists of relevant reviews were performed to identify additional studies.

Eligibility Criteria
This review considered studies that reported on health care provider attitudes, beliefs, experiences, and perspectives about prescribing opioids to manage CNCP or adhere to recommendations made by clinical practice guidelines. Studies conducted on health care professionals with privilege to prescribe opioids were eligible (e.g., physician, nurse practitioner, physician assistant), including those that reported on medical residents. Studies were excluded if perspectives of the health care professional and patient could not be segregated for synthesis.
Studies that used qualitative data collection and analysis methods and qualitative components of mixed methods studies were eligible, with conference abstracts and review articles excluded. Studies that were conducted in primary or specialist care within North America were eligible, given that opioid-related prescribing practices and nonmedical use and harms are greatest in this area of the world. 24 Studies were excluded if their primary focus pertained to (1) acute pain, (2) emergent medicine or prescribing within the emergency department, and (3) the management of problematic opioid use.

Screening and Selection
Two teams of reviewers independently screened titles and abstracts for eligibility and reviewed full-text publications of all potentially relevant articles using the reference management software Rayyan. 25 Disagreements on inclusion and exclusion of articles between reviewers were resolved by consensus or arbitration if necessary. Interrater agreement was quantified using Cohen's kappa and percentage agreement.

Approach to Evidence Synthesis
A metasynthesis approach was used to systematically identify and synthesize unique qualitative insights from health care professionals. Metasynthesis is a deductive and inductive process with an epistemological basis in objective realism and involves identifying findings, grouping findings into categories, and then grouping categories into synthesized findings. 26 Following the Thomas and Harden approach to qualitative evidence synthesis, 27 all text within primary studies was reviewed and coded independently by two reviewers (J.A.R., L.V. B.) to develop descriptive themes.
Meetings were held to review descriptive themes and generate analytical themes until saturation of themes was reached. Analytical themes were coded as barriers or facilitators to prescribing opioids for CNCP according to the theoretical domains framework (TDF). 28 The TDF distills 33 theories of behavior change into 14 domains that provide a theoretical lens through which to view cognitive, affective, social, and environmental influences on behavior. 28 Health care provider statements about barriers and facilitators were assigned to a relevant TDF domain, or set of domains, using a process described by Cane et al. 29 Two reviewers (S.F.F., L.V. B.) independently categorized barriers and facilitators. Discrepancies were resolved through consensus following meetings between team members. Barriers and facilitators of similarity were grouped into themes and agreed upon by consensus of three members (J.A.R., L. V.B., and S.F.F.). The result was a framework of 10 themes and 15 subthemes.

Quality Appraisal
Methodological limitations were appraised using the tenitem Critical Appraisal Skills Programme (CASP 30 ) tool. The CASP tool is the most commonly used tool in qualitative evidence syntheses in Cochrane and the World Health Organization guideline development processes. 31 Quality appraisal was performed in duplicate by three reviewers (L. V.B., J.A.R., and S.F.F.) who met and resolved discrepancies through discussion. Study quality was rated as low (met criteria for less than two domains), moderate (met criteria for two domains), and high (met criteria for three or more domains) depending on whether they met four core domains: (1) qualitative methodology was appropriate, (2) recruitment strategy was appropriate for research aims, (3) relationship between researcher and participants was considered, and (4) data analysis was sufficiently rigorous. These four domains were recommended as best capturing the strengths and limitations of qualitative research. 31

Assessing Confidence in Evidence
As recommended by the Cochrane Qualitative and Implementation Methods Group, 31 confidence in findings was assessed with the Grading of Recommendations, Assessment, Development, and Evaluation-Confidence in the Evidence from Qualitative Reviews (GRADE-CERQual) approach. 32 Four components were considered: (1) methodological limitations of included studies supporting a review finding, (2) the relevance of the included studies to the review finding, (3) coherence of the review finding, and (4) adequacy of the data contributing to a review result. Confidence ratings started at "high" and were downgraded if concerns were present.

Study Identification and Selection
We identified 12,253 citations and reviewed 383 fulltext articles. Interrater agreement was almost perfect at the title/abstract stage, k = 0.92 (95.98%) and moderate at the full-inclusion stage, k = 0.62 (80.78%). Twentyseven qualitative studies 33 − 59 met inclusion criteria. Two studies reported on the same data, 58,59 resulting in 26 unique studies. Figure 1 presents a flow diagram depicting inclusions and exclusions of citations reviewed. Characteristics of included studies are depicted in Table  1. Eligible studies enrolled 599 health care providers (487  primary care providers, 19 specialist physicians, 35 nurse  practitioners, 27 pharmacists, 13 physician assistants, 12 medical residents, and 6 reported as "other").

Methodological Quality Assessment
Supplemental Table 1 provides a summary of the critical appraisal of the quality of included studies. Overall, 6 studies were rated globally as having high quality, 10 of moderate quality, and 9 of low quality. Of note, the researchers' reflexivity (i.e., critical self-evaluation of the researcher's position and explicit recognition and acknowledgment of how this position may affect the research process and outcomes 60 ) was given insufficient documentation or consideration in a majority of studies (21 of 27). Also of note, almost half of included studies paid insufficient attention to the method, rationale, and discussion of the recruitment process (e.g., who elected not to participate and why) to contextualize the potential for selection bias (12 of 27). Table 2 depicts a summary of findings and Supplemental  Table 2 depicts a summary of qualitative findings and confidence assessments using the CERQual approach. Themes were categorized based on their impact on prescribing (i.e., barriers, facilitators, and mixed). This categorization was done at the level of theme. Results are presented in the following with representative quotes. A list of quotations supporting each theme is provided in Supplemental Table 3.

Subtheme 1.1: Inability to Perform an Unbiased Evaluation of Pain and Identify an Objective Explanation Decreased Providers' Willingness to Prescribe (GRADE-CERQual Confidence Level: High)
Nine articles [33][34][35]41,43,44,[51][52][53] reporting on 148 providers (90 primary care, 12 nurse practitioners, 12 residents, 18 internal medicine, 12 specialists, 1 health psychologist, 3 other) identified difficulty assessing pain in the absence of objective indicators, such as tissue damage. [24][25][26]31,33,34,[41][42][43] Providers noted that the subjectivity of pain complicates the ability to establish a diagnosis, leaving discomfort in managing pain and caring for the patient. In the words of one provider: "Somebody will say, 'Oh, I have ten out of [ten] pain', but they're sitting there, comfortably. They're walking around, totally fine, and you're like, 'This is not what I would consider ten out of ten pain.'" 44 Willingness to prescribe opioids also decreased when objective indicators of pain were not present: "If I can't find the cause of the pain, that is always an interesting problem. The reality is that I would never give somebody narcotics in that case." 43

Subtheme 1.2: Providers Who Questioned the Perceived Efficacy of Opioids Were Reluctant to Prescribe (GRADE-CERQual Confidence Level: Moderate)
Most providers working within primary care reported uncertainty about the efficacy of opioids for relief of pain and improvement of function. This theme was supported by five articles that reported on 81 providers (60 primary care, 7 nurse practitioners, 1 physician assistant, 12 residents, 1 health psychologist). 41,47,51,53,55 Some providers indicated that doubts over the efficacy of opioids arose from recent empirical evidence. Uncertainty about the efficacy of opioids was seen as contributing to a reluctance toward prescribing opioids to manage CNCP given the potential for adverse effects: Do we have evidence to show that these opiates are going to work in chronic nonmalignant pain? You know, no I don't think so. . . . The pendulum has kind of moved the other way. We're trying to be more thoughtful in prescribing. So I think that'll mean less prescriptions, stopping meds that are not effective or dangerous. Safety has become kind of an issue, a bigger issue as well. 47

Subtheme 2.1: Concern over Patient Safety Resulted in Caution When Prescribing Opioids (GRADE-CERQual Confidence Level: High)
Many providers wanted to "do the right thing" by helping their patients manage pain by prescribing opioids. However, patient safety was viewed as one of the most important factors when prescribing opioids, and the thought of potentially harming their patients (e.g., fear of causing addiction or overdose) was one of the most salient concerns that resulted in caution with regard to prescribing opioids for CNCP. This theme arose across 14 studies 34,35,37,39,41,[43][44][45][46][47]49,51,52,58 that reported on 252 providers (189 primary care, 27 nurse practitioners, 2 physician assistants, 10 residents, 8 internal medicine, 10 specialists, 2 osteopaths, 2 pharmacists, and 2 others). 2. Clinicians found it difficult to be on the receiving end of complaints regarding their perceived lack of concern for patients' pain, when they believed that their actions were ultimately in the patients' best interest (refer to theme 4) 3. Providers report that conversations about opioids were time consuming and appreciated that nurses provided group education visits so they do not have  Theme 8: Interprofessional support or mentorship improved provider willingness and confidence to engage in patient discussions surrounding opioid prescribing 48,53,59 Mentorship and support from specialists, experienced colleagues, or allied health professionals improved PCP willingness and confidence to engage patients in discussions around prescribing opioids for CNCP.
Low confidence Serious concerns about methodological limitations, adequacy, and relevance. Moderate concerns about coherence Theme 9: Use of PDMPs for risk management contributed to providers' confidence in opioid prescribing and discontinuation 33,35,36,39,40,47,49,50,52,54,55,58 The use of PDMP, UDTs, and patient history were perceived as beneficial to decision making for opioid prescribing, better management of medications, and the prevention of opioid abuse and diversion Moderate confidence Moderate concerns about methodological limitations and coherence Mixed Theme 10: External policies and tools increased providers' confidence in decisions to prescribe or discontinue opioids. Meanwhile, external pressures decreased willingness to prescribe 10.1 Providers adopted a defensive stance when prescribing opioids 33,34,40,41,43,44,46,47,52 Most providers adopt a position of distrust, or a "defensive stance," when entering a patient-provider relationship and prescribing opioids.  43,47,58,59 Providers reported that institutional policies around opioid prescribing and tapering set by an organizational opioid safety committee or ethics board reduced subjectivity and increased confidence in decision making Moderate confidence Moderate concerns about adequacy.
Serious concerns about methodological limitations Theme 11: Some providers struggled to enact guidelines, especially among complex patients. Contrarily, some providers felt the guidelines were helpful for making decisions and setting boundaries 37,50,52,55,58 Most providers felt that guidelines were difficult to enact and felt like they were "flying by the seat of their pants." For example, some felt like these guidelines were not straightforward or did not pertain to complex patients. Several studies within this theme discussed concerns over patient safety when tapering or discontinuing opioids, leading to the decision to continue prescribing while fully aware of and concerned about risks: "In others, stimulant use and alcohol use goes way up when I titrate down their opioids. So, prescribing opioids in a controlled fashion for their pain, despite their pain risk, seems to be less risky." 52 Some studies within this theme indicated that concern over patient safety was partially dependent on patient characteristics, such as age:

Subtheme 2.2: Concern over Drug Diversion Resulted in Caution When Prescribing Opioids (GRADE-CERQual Confidence Level: Moderate)
Seven articles including 134 providers (88 primary care, 17 nurse practitioners, 2 physician assistants, 12 internal medicine, 12 specialists) discussed concerns over drug diversion when prescribing opioids. [33][34][35]43,45,47,49 General practitioners felt that they were contributing to potential drug diversion within their community by prescribing opioids to their patients and reported reluctance to prescribe opioids as a result.
In my first year here, I was very kindly giving benzodiazopenes to a woman for neck spasms and Percocet for her neck pain based on her records and what she told me. And when I found out I was also prescribing for her sister and her mother I realized that single-handedly I was probably prescribing for all of New Haven and immediately got them off. 33 Concerns over drug diversion were complex and were influenced by the population that frequented the clinics where providers worked. As noted by one provider: "I think our population can divert quite a few meds. I think their financial situations can be really tenuous. Sometimes they sell pills to survive." 35 One provider indicated that the clinical decision-making algorithm favors harms to society under such circumstances: "The harm to society overrides the patient." 43

Subtheme 3.1: Negative Personal-Related Events Decreased Providers Willingness to Prescribe Opioids (GRADE-CERQual Confidence Level: Moderate)
Four articles elucidated the impact of negative salient personal events on opioid prescribing among a sample of 71 providers (45 primary care, 17 nurse practitioners, 2 residents, 2 osteopaths, 2 pharmacists, 1 health psychologist, 2 other). 39,43,53,58 Providers reported a reduced willingness to prescribe opioids for CNCP due to the experience of negative salient personal events. Some providers noted that the perceived blame for opioid-related patient overdose or death impacted their future prescribing practices. Others noted that they were not willing to prescribe opioids to their patients who were confrontational, threatening, or violent due to concerns over personal safety:

Subtheme 3.2: Negative Patient-Related Events Decreased Providers' Willingness to Prescribe Opioids (GRADE-CERQual Confidence Level: High)
Five articles discussed the impact of negative patientrelated events on prescribing opioids among a sample of 95 providers (78 primary health care, 7 nurse practitioners, 1 physician assistant, 8 internal medicine, 3 specialists, 2 other). 34,35,44,45,51 Negative patient-related events, such as patients misusing opioids or opioidrelated patient deaths, were associated with hesitancy when prescribing opioids. After recalling negative events, providers were more guarded about their prescribing behaviors in order to avoid future patientrelated overdose, misuse, or death.

Theme 4: Providers Expressed Difficulty Having Conversations Surrounding Opioid Prescribing and Feared These Conversations Would Harm the Therapeutic Alliance (GRADE-CERQual Confidence Level: Moderate)
Ten articles 33,34,36,40,46,47,49,52,53,58 highlighted the difficult nature of conversations surrounding opioids among a sample of 210 providers (152 primary care, 15 nurse practitioners, 10 physician assistants, 2 residents, 12 internal medicine, 12 specialists, 1 health psychologists, 6 other). Most general practitioners indicated that conversations surrounding opioid-related harms (e.g., side effects, misuse), tapering, and treatment agreements were difficult to broach with clients, fraught with tension, and avoided when possible. Some providers also found that these conversations ruptured the patientprovider relationship.
Those conversations around "I don't think prescribing this is appropriate" . . . physicians tend to shy away, because I think they expect them to be confrontational. . . . You want your patients to trust you, to believe you, you want the clinical encounter to be relatively conflict free. I mean, I think that's the biggest struggle, and I think some doctors, as a result, don't prescribe opioids. Which is wrong, because they're a legitimate clinical and pharmacological resource, but there's that fear. 52

Subtheme 5.1: Unavailability of Opioid Alternatives Facilitated Opioid Prescribing (GRADE-CERQual Confidence Level: Moderate)
Nine studies described the impact insufficient availability of nonopioid treatments for pain management has on prescribing opioids for CNCP among a sample of 215 providers (147 primary care, 12 nurse practitioners, 3 physician assistants, 10 internal medicine, 9 specialists, 2 osteopath, 27 pharmacists, 5 other). 33,39,47,49,52,54,55,57,59 Lack of available nonopioid treatments (e.g., specialty pain management services, community resources) was viewed as facilitating the prescription of opioids for CNCP. As noted by one provider: It takes some time to find a resource, for physio or for massage, all those other things that could help manage pain. And it takes some time to get those in place. There's not another great alternative pain management mechanism out there, both pharmacological and even, again, nonpharmacological. Often [patients don't qualify], and they can't get access to a lot of the other supports. 52 A perceived lack of alternative approaches to pain management was viewed as interacting with social factors, such as housing insecurity, lack of financial means or insurance, and poor access to transportation or social support, that further contributed to limited accessibility of care. Similarly, patient complexity (e.g., co-occurring substance misuse) and lack of funding to promote nonopioid therapies were viewed as further contributing to the problem.

Subtheme 5.2: Appointment Time Constraints Limited Thoughtful Prescribing Practices and Patient Education Surrounding Opioid Treatment (GRADE-CERQual Confidence Level: Moderate)
Information from 11 articles that included 212 providers (138 primary care; 7 nurse practitioners, 10 physician assistants, 2 residents, 16 internal medicine, 9 specialists, 1 health psychologist, 3 other) indicated that time constraints during appointments were a barrier for providers to adequately assess and manage pain. 33,40,44,46,49,52,53,[56][57][58][59] Providers noted that brief appointment times did not allow for meaningful conversations surrounding opioid prescribing, which led to inadequate patient education about pain and opioids.
And I think that is tough in our busy practices, to actually take time to really educate people about pain, and that's why we offer this chronic pain program through our family health team. . . . But you can imagine a physician that doesn't have access to that. You don't generally have the time to spend in sessions talking about how to manage pain. 52 We get our little ten-minute per patient, which is so grossly, woefully inadequate amount of time to see a patient. Ten minutes, right, for all your problems. And so nobody wants to take the time to explore things other than drugs for people with chronic pain. 49 The impact of short appointment durations on opioid prescribing was viewed as compounded by the volume of paperwork required when providers prescribe opioids. This led some providers to reduce the number of appointments accepted or limit the number of patients prescribed opioids on their caseload: That is why you have to limit the number [of patients prescribed opioids], you can't carry a lot of them because there's so much paperwork. 33 So, I heard all this stuff that I'm supposed to be doing, taking a complete history, complete addiction history. . . . But I don't have time to do what I am supposed to do in terms of proper treatment, opioid treatment, so I cut corners a bit. 56

Subtheme 5.3: Inadequate Formal Education during Medical School or Residency Resulted in Unstandardized Prescribing Practices (GRADE-CERQual Confidence Level: Moderate)
Four studies reported on the perceived adequacy of formal education surrounding the prescription of opioid analgesics for the management of CNCP among 114 health care providers (74 primary care, 12 nurse practitioners, 25 pharmacists, 3 other). 35,43,46,54 General practitioners indicated that perceived inadequate education on the prescription of opioids for pain management received during medical school or residency was a barrier to prescribing. As summarized by Kang et al. 54 : "Many participants felt that there was a lack of provider education on this topic." Some providers allowed personal experiences and interactions with clients to shape their prescribing habits, and others took the initiative to seek continuing medical training.
I have had a kidney stone, and if a person has a kidney stone, they get pain medication. 43 I've basically sought out my own training because it was not something that was taught to me in medical school or residency. 46

Theme 6: Prioritizing Improvement in Function Increased Providers' Willingness to Discuss Opioid Prescribing (GRADE-CERQual Confidence Level: Moderate)
Primary care providers prioritize improvement in function over reductions in self-reported pain and reported greater willingness to discuss opioid prescribing and discontinuation when the goal was to improve patient function. This theme was supported by four articles 41 37,41,46,50,52 Providers felt more confident prescribing opioids for CNCP when they perceived interactions with patients to be straightforward and honest with good two-way communication, highlighting the importance of cultivating a strong working alliance and a shared decisionmaking approach.
She is very straightforward and she hasn't caused any troubles-she hasn't broken any pain contracts. I have no reason to taper her off abruptly. 41 And the best thing about it is that I involve the patient when I do this, because it's not something secretive; it's out in the open. So, I'll say, "Okay, let's pull this out, let's print this [the pain contract] out." And in fact I've given patients the copies of this. 50

Subtheme 7.2: Long-Standing Patient-Provider Relationship and Trust Increased Willingness to Engage in Opioid Discussions (GRADE-CERQual Confidence Level: High)
Seven articles elucidated the impact of a long-standing patient-provider relationship on prescribing opioids among a sample of 195 providers (166 primary care, 4 nurse practitioners, 8 physician assistants, 8 internal medicine, 3 specialists, 2 osteopath, 2 pharmacists, 2 other). 34,39,40,42,44,50,52 Having a long-standing relationship with a patient increased willingness to engage in discussions surrounding opioid prescribing. Providers noted that knowing the patient well, knowing their background, having a well-rounded level of trust, and having a strong relationship with their patient increased their comfort level when prescribing opioids: "If you're the family doctor or long-term psychiatrist or internist who's known them [the patient] for 10 years and now they've developed a pain problem, you already have a bank of goodwill both ways." 34

Theme 8: Interprofessional Support or Mentorship Improved Providers' Willingness and Confidence to Engage in Patient Discussions Surrounding Opioid Prescribing Practices (GRADE-CERQual Confidence Level: Low)
Three articles that reported on 56 providers (46 primary care, 4 nurse practitioners, 2 physician assistants, 2 residents, 1 health psychologist, 1 other) explored the impact of mentorship on opioid prescribing. 48,53,59 Primary care providers indicated that mentorship and support from specialists, experienced colleagues, or allied health professionals improved willingness and confidence to engage patients in discussions around prescribing opioids for CNCP. Advice on how to approach challenging patient cases, support from pharmacists on how to design opioid tapering plans, and validation from colleagues on the challenges of managing patients with CNCP were perceived as particularly beneficial. [

Theme 9: Use of Prescription Drug Monitoring Programs for Risk Management Contributed to Providers' Confidence in Opioid Prescribing and Discontinuation (GRADE-CERQual Confidence Level: Moderate)
Twelve studies that included 271 providers (192 primary care, 16 nurse practitioner, 10 physician assistant, 10 internal medicine, 9 specialists, 2 osteopath, 27 pharmacists, 2 other) reported on the impact of risk management tools on prescribing opioids. 33,35,36,39,40,47,49,50,54,55,58 The availability of risk management tools for prescribing opioids, such as patient history, prescription drug monitoring programs (PDMPs), and urine drug testing, was perceived as beneficial and their implementation contributed to willingness to prescribe.
I had a patient who was getting oxycodone for six to eight months for a chronic pain issue . . . because we were concerned about his behavior, we did a urine tox which showed no oxycodone and, instead, showed morphine could've been heroin or anything he bought. So, we discontinued oxycodone. 33 Providers also noted that risk management tools helped to better manage medication and prevent opioid abuse and diversion. For example, providers felt that treatment agreements set clear expectations about opioid use that could lead to meaningful discussions when risk was present: "Okay, it's clear to us that you're not following through with the guidelines of the contract. And if that's the case, then . . . I don't feel comfortable prescribing for you anymore because you're using in a way that's unsafe." 58

Subtheme 10.1: Providers Adopted a Defensive Stance When Prescribing Opioids (GRADE-CERQual Confidence Level: Moderate)
Nine articles that included on 172 providers (101 primary care, 16 nurse practitioners, 9 physician assistants, 10 residents, 18 internal medicine, 12 specialist, 6 other) reported that it is advantageous for a provider to adopt a "defensive stance" (i.e., a position of distrust) when entering a patient-provider relationship that centers around prescribing opioids. 33,34,40,41,43,44,46,47,52 This defensive stance likely arises from vigilance toward the identification of "red flags" for the prescription of opioids, such as risk of addiction: "You become on guard right away. Whether rightfully or wrongfully, maybe to some extent, rightfully. But you right away think, okay, it's not just like my average, easy visit. This is going to be something that I've got to be more challenging and more aware." 52 Few providers 47 believe that it is their responsibility to be truly trusting of patient reports of pain. Rather, the majority of providers report skepticism or suspicion of patient motives, resulting in the adoption of a position of mistrust and a hesitancy to prescribe opioids.

Subtheme 10.2: Concern over Regulatory Scrutiny Decreased Willingness to Prescribe Opioids (GRADE-CERQual Confidence Level: Moderate)
Five articles reported on the impact that concern over regulatory scrutiny had on prescribing opioids among a sample of 70 providers (48 primary care, 16 nurse practitioners, 2 internal medicine, 3 specialists, 1 other). 34,43,45,50,51 Concern over regulatory scrutiny or being held liable for a patient's overdose or death led to a reduced willingness of general practitioners to prescribe opioids for CNCP.
If you have someone on a narcotic long term, it gets people's attention even if it is justified, so I feel like I am under the microscope when I prescribe those drugs. 43 There's always a lurking fear . . . certainly someone could overdose on what I prescribed and then their family member could try to press charges. 45 Some providers have witnessed colleagues lose their jobs over opioid prescribing and felt the need to protect their professional licensure, which led to a lower likelihood of prescribing opioids: Some providers expressed gratitude for the guidelines, indicating that they enabled helpful decision-making processes and helped set effective boundaries. As described by Chang et al. 37 : "Opioid prescribing guidelines enable helpful decision making processes when assessing risk, initiating opioids, or managing opioids. Guidelines help set effective boundaries (e.g., safe maximum dose, decision to taper or discontinue)."

Uncertainty about the Subjective Experience of Pain and Privileging Functional Outcomes
Pain is a complex and subjective experience that poses several measurement challenges. Currently, there exists no valid and reliable method of objectively quantifying the experience of pain. Uncertainty about the veracity of patient self-reports of pain was associated with diagnostic ambiguity and a reduced willingness to prescribe opioids. Until more objective physiologic/neurologic measurement techniques are perfected, clinicians who study pain will continue to rely on the careful use of established self-report measures of pain and its impacts. Privileging functional outcomes and active engagement in chronic pain self-management appeared to offset this ambiguity.

Concerns over Patient and Community Safety and Regulatory Scrutiny
Providers reported concerns over patient adverse effects, physical tolerance, and addiction. Moreover, experiencing or witnessing negative patient-related salient events (e.g., patients who have overdosed) and provider-related salient events (e.g., threat to provider) contributed to hesitancy surrounding prescribing opioids to manage pain. Salient negative events can result in cognitive biases that impact the delivery of medical care, 61,62 such as availability 63 and representativeness biases. 64 Cognitive behavioral techniques could be beneficial for highlighting cognitive biases.
For example, cognitive restructuring could be used to acknowledge distorted thoughts and promote reasoned practice. 65 Incorporating cognitive bias awareness into the curriculum at medical centers has yielded promising results, demonstrating that residents were able to recognize biases and create strategies to avoid biased reasoning. [66][67][68] Given associations between chronic opioid use and an increased risk for opioid use disorder, overdose, and death, [69][70][71][72] it is difficult to interpret whether concerns over adverse effects represent an accurate appreciation for potential opioid-related harms or relative risk aversion. Providers reported that concern over drug diversion was a barrier to prescribing opioids for CNCP. Population-based estimates indicate that the prevalence of diversion within the community is 5%, 73 with rates of diversion increasing as opioid prescribing rates increase. 74 Reasons for opioid diversion vary, ranging from unawareness of the consequences of opioid diversion, 75 to poor education about proper storage and disposal of opioids, 76 to aberrant medication behaviors or misuse. 77 Several strategies have been recommended to reduce drug diversion and identify those at risk of diverting, including (1) education, (2) urine screening, (3) patient contracts, (4) low dose initiation, (5) gradual dose titration, (6) use of PDMPs, and (7) completing multidimensional assessments with patients before prescribing opioids. [78][79][80] It is unclear whether these interventions have a direct impact on reducing opioid diversion. 78 Moderate confidence can be placed in the observation that concern over regulatory scrutiny was a barrier to prescribing opioids for CNCP. Though regulatory scrutiny and sanction can be effective methods for managing opioid prescribing behavior, regulatory policies vary by province, region, territory, state, and country and can result in barriers to prescribing, 81 as well as unintended consequences, such as patients not receiving necessary prescriptions and seeking illicit opioids. 82

Long-Standing Relationships with Patients, Trust, and the Working Alliance
Moderate confidence can be placed in the observations that the patient-provider working alliance was a robust facilitator of opioid prescribing and appeared to center around trust and safety. Providers reported that patients' withholding or minimizing information (e.g., history of substance use) reduced willingness to prescribe, 34,36,40 whereas open communication and a trusting relationship facilitated opioid prescribing. 39,41 Long-standing patient-provider relationships and shared decision making can reduce the impact of this complexity.
Research suggests that physicians demonstrating compassion, having a strong working alliance, and portraying a patient-centered approach that involves patients in treatment decision making improve provider clinical decision making, positive communication, patient health, and treatment adherence. 83,84 Punwasi et al. also observed that long-standing patient-provider relationships increase general practitioners' confidence in prescription decisions. 85 The adoption of such an empathic therapeutic approach may be particularly pertinent when prescribing opioids given that providers found conversations surrounding opioids to be difficult and potentially detrimental to the working alliance.

Education, Time Constrains, and Availability of Nonopioid Alternatives
Moderate confidence can be placed in the observation that primary care providers are reluctant to prescribe opioids for CNCP due to insufficient formal education on opioid prescribing. Canadian medical undergraduate education programs offered an average of 16 h on pain management in 2007, 86 and estimates suggest that only 20% of physicians in the United States received training about recognizing drug diversion or identifying signs of substance use disorders while in medical school. 87 Health care providers also reported a lack of confidence in their ability to prescribe opioids to manage CNCP, specifically when no medical explanation for pain is present 35,51 and in the presence of comorbidities (e.g., substance use disorder or mental illness). 33,48 Education and training in chronic pain are necessary to enhance provider knowledge, confidence, and self-efficacy surrounding chronic pain management, opioid prescribing, and identifying drug abuse and addiction among people with chronic pain. 88,89 Moderate confidence can be placed in the observation that a lack of readily accessible nonopioid alternatives (e. g., referral pathways, community resources) was a facilitator to prescribing opioids. A similar theme was extracted from a previous synthesis of qualitative literature on prescribing opioids for chronic pain. 85 Insufficient nonopioid alternative treatments can result in protracted wait times, which is problematic given that wait times for chronic pain care in excess of 6 months have been deemed medically unacceptable. 90 As evidence for this, more than 50% of patients referred to chronic pain clinics in Canada are not admitted within the recommended wait time, with many waiting years for admittance, a circumstance that has not changed over the last 10 years. 91 Canadian guidelines for opioid therapy in CNCP highly recommend maximizing the use of nonopioid therapies. 12 Yet, access to such therapies is difficult due to long wait list times, few multidisciplinary support programs available, 91 a lack of digital platforms for chronic pain self-management, 92 and lack of access to publicly funded allied health professionals (e.g., physiotherapy, psychology). 93 Moderate confidence can be placed in the observation that brief appointment durations are a barrier to prescribing opioids for pain management. The management of chronic diseases often requires more time than physicians have available, which is a significant obstacle to the delivery of quality care. 94 This is interesting given that a recent cross-sectional study of physician behavior in primary care reported that opioid prescriptions increased by 33% as the workday progressed and by 17% when appointments ran behind schedule. 95 It would appear that brief appointments in combination with a perceived lack of availability to discuss nonopioid therapies influence clinical decision making when prescribing opioids.

Relationship between the Current Synthesis and the COVID-19 Pandemic
It is important to note that we synthesized evidence published before the global COVID-19 pandemic. The COVID-19 pandemic had a clear impact on pain management globally, including prescribing opioids for CNCP. The COVID-19 pandemic introduced additional considerations for prescribers when weighing the potential risks and benefits associated with prescribing opioids (e.g., opioids can increase the risk of respiratory depression, suppress the immune system, or mask fever and myalgias that may be related to . 96,97 Further, providers were forced to transition to telemedicine, which imposed additional challenges to assessment and maintaining strong therapeutic relationships. 96,97 Unfortunately, there is insufficient evidence to appreciate the full impact of the COVID-19 pandemic on the prescription of opioids for CNCP or whether the impact will be transient or long-lasting. We contend that a strong appreciation of the barriers and facilitators for prescribing opioids for CNCP before the COVID-19 pandemic will aid in our understanding of how prescribing practices have evolved during and after the pandemic.

Strengths of This Review and Reflexivity
This review was undertaken by a multidisciplinary team with diverse viewpoints and experiences in chronic pain management, including two practicing clinical psychologists with expertise in chronic pain management, one practicing anesthesiologist with expertise in chronic pain management, and two trainees (one experimental and one clinical) from the discipline of psychology.
Members of our team engage in regular clinical practice and have clinical experience working with people who live with pain and use opioids with benefits and opioid-related harms. Members of our team are involved in federal mentorship programs for chronic pain, substance use, and addictions and co-developed the safe prescribing course for opioids. As such, we likely tended to identify and code potentially modifiable barriers and facilitators that coincide with evidence and theory on the adoption of recommendations made by clinical practice guidelines. It is also important to note that we coded barriers and facilitators based on the likelihood of decreasing or increasing the frequency of an event, respectively. There was no value judgment placed on the appropriateness of such deviations to clinical practice. This is one of the first reviews to qualitatively identify health care providers' perceived barriers and facilitators when prescribing opioids for patients with CNCP. This synthesis will serve as a starting point when understanding potential changes in barriers and facilitators of opioid prescribing that may have occurred since the COVID-19 pandemic.

Limitations of This Review
First, we focused on evidence from North America. The results of this synthesis may not extend to other geographic regions given the sociopolitical nature of the opioid crisis. Second, studies that focused on opioid use disorder and the prescription of opioid agonist and antagonist treatments were excluded. Thus, results are not generalizable to the prescription of opioid agonist treatments. Third, this review focused on prescribing opioids rather than the alignment of prescribing with practice guidelines due to a lack of available literature. As such, some facilitators may represent barriers to appropriate prescribing. For example, a lack of opioid alternatives facilitates opioid prescribing yet can be a barrier to optimizing nonpharmacological care prior to prescribing opioids for CNCP.

Conclusion
Providers have an ethical responsibility to work with patients to manage CNCP, and such management may include chronic opioid therapy. The current synthesis identified barriers to prescribing opioids, including providers' uncertainty regarding the subjectivity of pain and efficacy of opioids, concern for patient and community safety, past negative experiences, and fear of harming the therapeutic alliance. Facilitators to prescribing opioids included prioritizing improvement in patient function, a strong therapeutic alliance, and interprofessional mentorship and support. Our findings also revealed complex processes that could act as barriers or facilitators to prescribing opioids for CNCP, such as external policies, opioid prescribing guidelines, and organizational constraints. Suggestions for improving judicious opioid prescribing include continuing education, mentoring or support from experts, and ensuring that regulators apply scrutiny consistent with best evidence.