‘Do they care?’: a qualitative examination of patient perspectives on primary care clinician communication related to opioids in the USA

Abstract

Objectives

This substudy’s objectives were to (1) examine the transferability of a four archetype framework (simplified pattern of prototypical features) for patients at high risk for opioid use disorder (OUD) developed from a previous study with a similar population; (2) explore how patient preferences for terminology can inform clinician communication strategies for patients with OUD across archetypes and (3) explore how patient perceptions of opioid risks can inform clinician communication strategies across patient archetypes.

Design

This qualitative study collected data via semistructured phone interviews with patients about views on opioid-related discussions with primary care clinicians. Qualitative data were coded using the Rigorous and Accelerated Data Reduction technique and analysed via iterative inductive/deductive thematic analysis.

Setting

40 primary care clinics affiliated with two health systems (site1=Pennsylvania; site2=Minnesota, Wisconsin and North Dakota).

Participants

40 adults meeting one of the following: OUD diagnosis; taking medication for OUD (MOUD) and ≥3 opioid prescriptions in the previous year.

Results

The aforementioned four archetype framework transferred well to the study sample and hinted at archetype differences in participant OUD-terminology preferences and opioid risk perceptions. Two additional archetypes of ‘in treatment/recovery for OUD and not taking MOUD’ and ‘in treatment/recovery for OUD and taking MOUD’ were identified. Participants best fitting archetypes 1–4 preferred clinicians to refrain from using addiction terminology to describe their relationship with opioids, finding the term ‘dependence’ as more appropriate and a signal that clinicians cared for patients. Participants who best first archetypes 5–6 felt ‘addiction’ was an appropriate, direct term that accurately described their condition, often using it themselves. Patients in all archetypes recognised risks of harm from using opioids, especially participants fitting archetypes 2, 5 and 6 who conveyed the greatest concern.

Conclusion

The modified six archetype framework may help clinicians tailor their communication and care for patients diagnosed with or at high risk for OUD.

Trial registration number

NCT04198428.

Correspondence to Dr Anthony W Olson; [email protected] ; Dr Katrina M Romagnoli; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• The qualitative methodology with a large and diverse patient population enables an in-depth exploration of the subjective realities experienced by individuals diagnosed with or at high risk for opioid use disorder.

• Robust and systematic approach to collecting and analysing qualitative data collected from two separate health systems in geographically dispersed states.

• The findings of this study may not be generalisable to healthcare organisations in urban areas, non-white populations and other countries.

Background

Opioids have contributed to the deaths of roughly 1 million Americans since 1999, with an estimated 84 000 deaths in 2023.^{1 2} Opioid overprescribing in the 1990s ignited an epidemic and over time led to more misuse and illicit use of opioids among individuals who went on to develop opioid use disorder (OUD).^{3 4} There were 3 million Americans living with OUD in 2021, a disease that is likely underdiagnosed, and over three times that many who misused opioids.^4–6

The Centers for Disease Control and Prevention Clinical Practice Guideline for Prescribing Opioids for Pain advises that clinicians regularly assess and discuss opioid risks, benefits and goals for treatment with their patients.⁷ Primary care clinicians (PCCs) face several challenges in identifying and supporting patients at risk for OUD, including appropriately selecting which patients to screen, complex screening tools, staff hesitancies and discomfort, workflow impediments that reduce follow-ups, institutional memory loss from staffing shortages and turnover, low screening yields and stigma surrounding OUD.⁸ Many of these obstacles can be mitigated with the use of clinical decision support (CDS) in electronic health records (EHRs),^9–13 which provide ‘timely information, usually at the point of care, to help inform decisions about a patient’s care…(that) help clinical teams by taking over some routine tasks, warning of potential problems, or providing suggestions for the clinical team and patient to consider’.¹⁰ Uses of CDS in the care of patients using opioids or at high risk for OUD include prominently displaying criteria for making OUD diagnoses, order sets for labs or medications like buprenorphine, recommendations for referrals or other preventative care measures and notifications of potentially harmful situations.

However, patient–clinician discussions about opioids and OUD are sensitive, complex and can be affected by the strength and quality of their relationships.^14–17 Clinicians approach discussing opioid use, risk and treatment for abuse with patients in varying ways^{18 19} and may use different communication styles for each unique patient (eg, motivational interviewing informed by the transtheoretical model (TTM) of health behaviour change).^20–23 Hooker et al ²⁴ examined patient perceptions of opioid risk discussions with PCCs prompted by an OUD-CDS and identified four archetypes of patient relationships with opioids: (1) use of opioids for chronic pain; (2) use of opioids for acute pain; (3) problematic opioid use and not open to treatment and (4) problematic opioid use and open to treatment. The archetypes represent a simplified pattern of features at one point in time that can help clinicians effectively initiate conversations involving opioids, opioid risks and OUD. Importantly, the archetypes are not definitive, permanent or algorithmic categories that define patients.

The qualitative substudy presented in this article was preceded by Hooker et al’s in the same multicentre pragmatic clinical trial investigating the effectiveness of an OUD-CDS designed to support PCCs’ care for patients with or at high risk for OUD.^{13 24} The overarching trial’s design was characterised by a staggered start for the three health systems involved (HealthPartners, Geisinger, Essentia Health; see online supplemental material for demographic comparisons) using uniform interventions and eligibility criteria, enabling sites to learn implementation lessons from each other as the overarching trial progressed. The objectives of our substudy were to (1) examine the transferability of Hooker et al’s four archetypes of patients with OUD to a distinct but similar population; (2) explore how patient preferences for terminology can inform clinician communication strategies with patients about opioids and OUD across archetypes and (3) explore how patient perceptions of opioid risks can inform clinician communication strategies with patients across archetypes.

Methods

This preplanned, experiential phenomenological mixed-methods substudy was conducted at 40 PCCs in the Geisinger (site 1; Pennsylvania) and Essentia Health (site 2; Minnesota, Wisconsin, North Dakota) systems. These health systems were staggered second and third in the overarching multisite clinical trial described in the background. This substudy is a follow-up to Hooker et al’s preceding substudy conducted solely at HealthPartners (Minnesota) primary clinics.

Study eligibility and recruitment

Patients eligible for study inclusion were adults between 18 and 75 years old who either had a diagnosis of OUD, prescribed medications for OUD (MOUD), or received three or more prescriptions for opioids in the prior year, which is known to increase risk of developing OUD.²⁵ Patients with a diagnosis of OUD were identified using International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes for OUD and/or prescribed MOUD. Patients also must have seen a PCC at a trial clinic in the 6 months before being invited to participate in the substudy. Participant eligibility was first screened using EHR data and then confirmed with participant self-reports in the recruitment process. Study-ineligible patients were those who were flagged by the OUD-CDS as having active cancer diagnoses in the last year (except for non-melanomatous skin cancer) or in hospice. Purposive sampling was used to maximise the trustworthiness of the findings.^26–28

300 eligible patients were invited (email: n=176; EHR patient portal message: n=110; letter: n=14) to participate in the substudy between 25 June 2021 and 13 August 2021 at site 1 and between 1 November 2021 and 1 March 2022 at site 2, and at least one additional contact was attempted for all non-responders. Patients who initially responded but did not schedule or complete interviews were lost to follow-up without providing a reason for dropping out. Recruitment ceased when data saturation was independently reached at each health system.²⁹

Data collection

Data were collected with semistructured phone interviews conducted by AB (male), AWO (male), CMG (female), CIA (male), KMR (female), LAF (female) and LDT (female) between 12 July 2021 and 24 August 2021 at site 1, and between 7 December 2021 and 8 March 2022 at site 2. All data used for analysis were based on participant self-reports and were not cross-checked with EHR data (eg, chronic pain diagnosis, comorbidities). The average duration of interviews was 51 min (range: 19–84 min). Two interviewers were on the call for each interview, taking field notes during the sessions and discussing their observations immediately after the interview. Interviewers did not have an established relationship with any study participants prior to study commencement and introduced themselves and the interest and goals behind the study at the outset of the interview. Interviews were audio recorded and then transcribed by HIPAA-compliant means.³⁰ Interview guides were pilot-tested with the research team and can be found in online supplemental material. Interview stem-question topics were (1) the patient’s relationship with their PCC, (2) the patient’s experiences taking opioids, (3) past discussions the patient had with their PCC about opioids and their perceived importance of those discussions, (4) past discussions the patient had with their PCC at appointments for chief complaints not related to opioids, (5) the patient’s perceptions and preferences for OUD-related terminology, (6) the patient’s reactions to OUD-CDS generated handouts containing targeted messaging and (7) the patient’s advice for PCCs when discussing opioid risks.

Coding procedure

Two experienced coders (AWO and KMR) trained five substudy team members (AB, CIA, LDT, HAH-B and CMG) in coding and analysis procedures. First, all coders conducted a preliminary review of the transcripts using the Rigorous and Accelerated Data Reduction (RADaR) technique to become familiar with the data and inductively generate initial codes.³¹ RADaR is a rapid but rigorous analytical approach ideal for time-constrained applied research. Each continuous statement by a speaker was the unit of analysis and multiple codes could be applied to each unit.

Second, a codebook was developed (by AWO and HAH-B) that hierarchically organised, merged and created new codes through an iterative inductive/deductive process informed by interview guide topics. The codebook transferred codes from the related qualitative study at HealthPartners that was conducted earlier in the trial²⁴ and combined them with the inductively generated codes from the RADaR technique. The codebook was then broken into three topical sections by AWO and KMR, with each coder being assigned one section of the codebook to use for analysis. Each codebook section was assigned to at least two coders.

Third, all transcription files were converted by AB into Microsoft Excel using a structure and process developed at Vanderbilt University.³²

Fourth, two randomly selected transcripts were independently analysed by all coders. Each coder deductively applied existing codes and inductively generated new codes when appropriate for their assigned sections. Coders covering the same codebook section compared their findings and resolved discrepancies through discussion.

Fifth, the previous step was repeated by each codebook section group until an a priori threshold of 80% interjudge coding reliability was met (agreement threshold before resolving discrepancies in each round). The targeted interjudge coding reliability threshold was met by all coders within completing four rounds. Then all remaining uncoded transcripts were divided equally among the seven team members and coded independently. After codes were applied to the transcripts using Excel templates, the coding was replicated exactly in Atlas.ti by AB and checked for accuracy by KMR.

Thematic analysis

Using the code frequency and co-occurrence analytical tools in Atlas.ti, the data were reviewed by KMR and AWO to identify emergent qualitative themes, using the archetypes of participants’ self-described opioid use as a framework for analysis. Analysis memos based on the Atlas.ti analyses were written regarding each theme and participant responses which were not represented by the existing four archetype framework. The memos were discussed by KMR and AWO on an ongoing basis for clarity and accuracy. After several iterations, consensus was reached and participants’ stories were sorted into the final categories which evolved into the six archetypes.

Patient and public involvement

Neither patients nor the public were involved in the design, conduct, reporting or dissemination plans of this research.

Results

68 patients responded to the invite, 53 scheduled interviews and 40 patients completed an interview (no repeat interviews were conducted). 40 patients completed interviews, 26 from site 1 and 14 from site 2. The majority of participants were female (68%) with an average age of 51 and a range between 24 and 75 years. All participants self-classified as white, with one person also identifying as Hispanic or Latino (3%). Most self-reported current opioid use (63%) and/or an OUD diagnosis (58%). Additionally, 35% self-reported current MOUD use and 28% self-reported present or past chronic pain.

Archetypes of self-described opioid use

Table 1 identifies and differentiates six archetypes for patients at high risk for OUD: (1) use of opioids for chronic pain; (2) use of opioids for acute pain; (3) problematic opioid use and open to treatment; (4) problematic opioid use and not open to treatment; (5) in treatment/recovery and not taking an MOUD and (6) in treatment/recovery and taking an MOUD. Four of these archetypes were previously identified in our colleagues’ work on the same topic. However, one characteristic of archetype 3 differed with several participants in our study sample perceiving opioids as necessary to manage pain. Hooker et al’s sample reported opioids were not seen as necessary to manage pain.²⁴ Archetypes 5 and 6 (in treatment/recovery not taking an MOUD and in treatment/recovery taking an MOUD) are newly identified archetypes that are described further below. The addition of these archetypes also resulted in three additional differentiating characteristics of ‘evidence of or confirmed OUD’, ‘in treatment/recovery’ and ‘taking an MOUD’ to Hooker et al’s original four characteristics. The total number of participants in each archetype exceeds the total number of participants because their stories often described how their relationship with opioids and perception of risks evolved over time. Thus, the same participant might be counted in multiple archetypes that best fit different points of their lives.

Table 1

Archetypes for patients at high risk for opioid use disorder, N=40 (adapted from Hooker et al ²⁴ )

Archetype	Long-term opioid use	See opioids as necessary to manage pain	Self-perceived risk of problems with opioids	Evidence of or confirmed OUD*	Open to treatment for OUD	In treatment or recovery*	Taking an MOUD*
People who use opioids for chronic pain (n=11)	Yes	Yes	No	No*	No	No*	No*
Exemplar quotes “As I said, I don’t [have concerns about the risks of opioids], because I write all of that down. I know I'm not taking more than I should and I know I can spread it out farther. I know my body is probably used to them, because I've been on them for so long. I feel that you know if I could, if I can bear the pain, I could go off of but I know I can't because I've tried and tried other things and I just can't do it. I can't function throughout the day like I said. I see (sic) stuck in bed and just end up dying in bed. I can't do that now.” 65+yo, F, Site 1 “I take it because I can't function without it. My mind is not addicted to this drug. My body probably is. And if I went off of it, I would probably suffer both pain-wise and withdrawal-wise.” 65+yo, F, Site 2
People who use opioids for acute pain (n=21)	No	Yes	No	No*	No	No*	No*
Exemplar Quotes “I think they did give me some [opioids], but I'm always afraid I'm going to get addicted, so I stopped taking it even though I was in terrible pain.” 55-64yo, F, Site 1 “No, I never worried about overdosing, but I worried about not being able to get off [opioids]. I just took them as they were prescribed.” 65+yo, M, Site 2
People with problematic opioid use who are not yet open to treatment (n=9)	Yes	Varied*	No	Yes*	No	No*	No*
Exemplar quotes “We worked in Europe a lot so I would have eight- or nine hour flights to Europe, and the pain was unbearable. So that was kind of my first—so [former PCP name] would give me hydrocodone just for those transatlantic flights, which really helped. So that’s kind of my first experience with opioids. And then the other time was I had a total left knee replacement in [date]… But when it ran out, I realized that they didn’t want to give me anymore. And I still couldn’t sleep at night. I was in horrible pain. But I also realized that I was up against the medical profession trying to right the ship and be more careful with them. But at that point, I was begging for them. And not because I was addicted, but because I was in so much fucking pain. Pardon me. So then you as a patient, I feel like I’m a criminal. People think I’m a criminal because I’m in pain.” 65+yo, M, Site 2 “Well, my opinion about that is—like you said, it’s coming from an addiction background to other things. If you are on Suboxone or methadone, you're already trying to curb it with a medical means of just another drug, which is just as bad as an opiate. To me that’s the same. It’s one and the same, and it’s not used right. It’s to be used for a short term, and you get off of it. It’s to help you with withdrawals to be brought down off of it. And the providers—I don't know if it’s just in [state name] or if it’s everywhere, but they never take you off of it. They will—for like if you are coming off it for drugs you could go to a methadone clinic or get a [MOUD] script, whether it be—I've even seen orders through the judge at cases that you get put on that, but they don't ever take them up. So why are you better than somebody, because you're on the Suboxone or methadone, which I almost feel is worse than being on a opiate—how they make you feel—what they do to you. Like I said, I was on methadone at that one time with those other medications, and I know how it feels and what it does. So to me, telling somebody who—you're just kind of like, “Oh, you're fine being on it forever.” You're still an addict. An addict is an addict; a drug is a drug. That’s how I personally feel about it. I don't like them. I mean, maybe I'll have to be on sometimes to get off of these or whatever—I don't know—or we wean them down. That’s what we did last time when I was on them is I had to medically step down.” 35-44yo, F, Site 2
People with problematic opioid use who are open to treatment (n=14)	Yes	No	Yes	Yes*	Yes	No*	No*
Exemplar quotes “It was getting so anxious and having to be on it so much, it was just consuming my life. And it was really hard to be in that much pain and not understand. Because I also didn't even understand that I was on these medications or what they even did to my body.” 25-34yo, F, Site 1 “When I was getting the prescription, is when the big opioid pandemic was going on. And people were dying, people were overdosing, and then it finally started hitting the news. And I was like, “Holy cow, that’s what’s happening to me.” And so they showed me medications that I could use just in case anything were to ever happen to me. They broke down that in my Suboxone, that there is medication already in there that help(s)it so nothing will happen to you. Thankfully, nothing bad. You can’t take too much or anything seriously bad [can] go wrong. And it was a huge, huge lifesaver because now I wasn’t feeling all that pain anymore and also wasn’t having any symptoms, like I was, anymore.” 25-34yo, F, Site 2
People who are in treatment/recovery for OUD and are not taking MOUD* (n=8)	Yes	No	Yes	Yes*	Yes	Yes*	No*
Exemplar Quotes “I just didn’t I just wanted to get back to being normal again. I don't want any substances in my body really. I could feel the haziness and the unclear decision-making [when taking a MOUD]. Even though [you are] not high off of [an MOUD], you feel affected by it. Your eyes are pinned out and just look like you are on something and my wife couldn't stand seeing me that way either, she would constantly nag me, you look you are high, but really I was just taking my [MOUD].” 35-44yo, M, Site 1 “Well, I didn't want to ever overdose. I always kept track of how many hours I would take them, as prescribed, and I knew that if I took too many, I could get addicted, which I think I kind of was anyway. And the risk was either [to] be pain-free or die from taking too many. So I stopped taking them. And I'm paying for it because I'm walking really poorly now. But I have a happier life.” 65+yo, M, Site 2
People who are in treatment/recovery for OUD and taking MOUD* (n=7)	Yes	No	Yes	Yes*	Yes	Yes*	Yes*
Exemplar Quotes “It took a lot of work to finally get into a place. I got into a place and told them all what was going on and everybody was really great and really supportive. I started taking [MOUD] … and I've been taking it ever since. I have not taken any kind of opiates nor needed them…It’s been really, really helpful. I do have concerns about being dependent on that, but my doctor and I that prescribed it, he asked me about that and what I thought and I just said I don't have an answer for what my long-term plan is. I just don't want to be in this situation anymore.” 35-44yo, F, Site 1 “I quit taking [opioids]. I stopped taking anything, it was about [date], and then I stopped taking… I was on benzodiazepines too, which are controlled too, but I was able to stop them in November. I weaned down over like 3 months with that, but the [MOUD] also helped me do that without withdrawal, which I can’t believe how good my anxiety has been since I quit taking them. As soon as that stuff would wear off, I’d have to take one or two more, and they had me on that for 11 years alone, just three of them a day, 3 mg a day, and I can’t believe they kept me on it that long, but they just sold it, so I wasn’t gonna not do it (sic), you know what I mean, being an addict, and thinking I really needed the medication until I actually stopped it and realized my side effects were coming from not having the medication, but once I was finally able to realize that, I was able to stop everything.” 25-34yo, F, Site 1

The same participant may have given exemplar quotes of different archetypes because their relationship with opioids evolved over time.

*Finding differed from Hooker et al ²⁴

Participant perceptions and preferences of terminology

Participants identified terminology used by clinicians that influenced the participant’s experience of discussing opioids with clinicians. A majority of participants in types 1–4 expressed that the terms ‘addiction’, ‘addicted’ and ‘addict’ hold a negative connotation (table 2). They preferred not to be referred to or have their condition or diagnosis referred to using any form of the word ‘addict’. Conversely, participants who described themselves as being in treatment/recovery from opioid addiction (archetypes 5 and 6) felt ‘addiction’ was an appropriate, direct term that accurately described their condition, often using it to describe themselves. However, they acknowledged it does have a negative connotation and might offend or upset others, particularly those who are experiencing addiction but are not ready to admit it to themselves. Participants also viewed addiction as a more nuanced term than dependence, which is reflected in the greater number of addiction-related codes used in the analysis (table 2).

Table 2

Terminology preferences

	Individuals who mentioned (n=40)	Positive mentions	Negative mentions
‘Dependent’ (n=65)	27	52%	26%
Exemplar quotes “Yeah. I think dependency is a pretty good one. I mean, if you're just starting to talk to someone about it, I don't always agree with just jumping in and nailing someone to the cross because that’s a turnoff to most people, I would say.” 65+yo, M, Site 2 “(Regarding the use of the word "Dependent)Yep. I feel like that’s easier to hear from my standpoint, where I'm totally dependent on my meds. I would never make it a day without them right now due to pain. But I don't feel like it’s an addiction because I'm taking them because I need them. And I feel like people are more receptive to softer words and not such accusatory language such as “addiction” and “addict,” especially when a lot of us have gone through hell trying to get our pain diagnosed and managed.” 25-34yo, F, Site 2 "I would think dependency. I wouldn't think that they were automatically doing anything wrong because it builds up in your system, and I know from my own experience, lowering the medication, especially if you don't have anything to take in the meantime, is kind of the way your body is going to react to it and your pain level and everything, and I would say that dependency is not necessarily the person doing anything wrong. It’s just how your body is reacting to what you've been taking for so long.” 45-54yo, F, Site 1
‘Addiction’ (n=146)	33	16%	37%
Exemplar quotes “Addiction is a powerful word. For me, it has an awful lot of meaning because, like I said, I understand where it comes from and what it is. Some people might be offended by the word addiction because the shortened version of addict, they don't want to consider themselves an addict to prescribed medication. I don't have an issue with that. I tell people I'm addicted to my high blood pressure medication also. It allows me to function during the day, as well as the opioid and everything else that I take. Yes. So, yeah, in those terms, it’s neutral to me. Addiction is what it is. It’s just a state of being that you truly don't want to be in.” 55-64yo, M, Site 1
“Addiction” means abuse (n=38)	21	21%	29%
Exemplar quotes “[Addiction] means that you're not taking it as directed, that you're abusing it.” 35-44yo, F, Site 1 “[Addiction means) that you're taking medication or illegal drugs and you can't stop yourself.” 55-64yo, F, Site 1
Use of ‘addiction’ appropriate in primary care (n=25)	13	48%	32%
Exemplar quotes “I don’t know, addiction, just a way that I think and act. It’s the way my brain is wired. That’s what I think when I think of addiction.” 25-34yo, M, Site 1 “I didn't like [the term addiction] when I was younger, but I came to grips with it through treatment, realizing and just coming to terms with the fact that to me I am an addict. I always will be an addict. I'm just an addict that’s (sic) in remission, and I think it’s just being straightforward and blunt. Addiction is the problem. It’s the best describer of the issue. And really, I don't take any—using the word or being around the word at all doesn't bother me, but that’s only because I've accepted it.” 35-44yo, M, Site 2
‘Addiction’ reflects clinician’s failure (n=10)	6	20%	90%
Exemplar quotes “I feel like addicts—a lot of times, those that become addicts because of a physical ailment or physical problems were somehow let down by their physicians not doing what they needed to do.” 45-54yo, F, Site 2 “Once the pain was gone and the prescriptions ran out, I was struggling a little bit after that. I actually felt like I needed more. I think because of the age that I was, I definitely got addicted to the [opioids], but again, like I said, [previous hospital name], they're a for-profit hospital. Back when that happened, we weren't really in an opioid crisis as bad as we are now.” 25-34yo, M, Site 1
‘Addiction’ is a disease (n=13)	7	23%	15%
Exemplar Quotes “I’m really not 100% ashamed of it because I know it’s a disease and that I’m not just a bad person.” 55-64yo, F, Site 1 “I do know that people who are prone to addiction, that when taking this medication some people feel the urge to take more than they're supposed to, they run risks of overdose or even addiction withdrawal symptoms when not taking medication.” 18-24yo, M, Site 1
‘Addiction’ has a negative connotation (n=42)	27	29%	69%
Exemplar quotes “But I guess it’s been—addiction or the word addict has just been used as such a negative term for so long that I feel like it’s a word that’s almost been kind of ruined, that I just wish there was (sic) a little more people knowing that this is a real health problem. It’s just kind of like any other health problems (sic) that you need to treat, that it’s not always a horrible person behind it.” 25-34yo, F, Site 2 “Well, I think addiction is a really useful term when talking about tolerance and withdrawal and the signs of addiction, but also it’s an important word. So it’s a harsh word and it’s full of maybe stigma, but it’s real. It’s true. And that’s a huge problem, right, with what we're talking about. So yeah, I much prefer that word.” 65+yo, M, Site 2
‘Addiction’ means different things to different people (n=22)	17	14%	36%
Exemplar quotes “I feel like addiction is a more comfortable word than like opioid problem…I think addiction just covers a wide range of things, not just one. I don't know. It’s just kind of narrowing one specific thing and addiction covers anything. Like you have more people reach out just in general. Because that one is just about opioid problem.” 25-34yo, M, Site 1 “Addiction is pretty wide open. It’s across the board. It’s not just one. I mean, addiction can be used in so many different situations. It can be used in alcohol. It can be used for gambling and all that stuff. So it’s not just one situation.” 35-44yo, F, Site 2

Positive or negative mentions refer to the percentage of instances in which a code from the first column was applied simultaneously with either the positive affect or negative affect code. Affect refers to the emotional meaning conveyed through the participants’ responses.

As participants would switch between discussing their previous and current behaviour and perceptions of opioids, multiple archetypes were identified in the same participant. The sum of participants in all archetypes exceeds the total number of participants. Archetypes 1–4 transferred from Hooker et al. ²⁴

Some exemplar quotes represent participant opinions at odds with existing knowledge and evidence. For example, while suboxone offer some protection for overdose it does not eliminate the risk and is intended as a long-term treatment rather than a short-term one.

MOUD, medications for opioid use disorder; OUD, opioid use disorder.

Most participants in non-treatment/recovery archetypes, preferred the terms ‘dependency’ or ‘dependent’ to describe someone having a hard time with opioids, because it lacked the negative connotation the word ‘addiction’ has. Participants also indicated it was more accurate to describe someone who needs opioids to manage significant pain and live a relatively normal life without implying they are using opioids deliberately to get high. They preferred clinicians use ‘dependency’ instead of ‘addiction’ when discussing opioid use with patients as it does not have the same accusatory language built into it which can negatively affect clinician–patient communication. For some participants, the use of ‘dependency’ was one type of signal that the clinician cared about them as a person and considered all the aspects of their health and lives when making treatment decisions. In the eyes of several patients, such an expression was evidence of a clinician using a biopsychosocial approach,^{33 34} which holistically considers the complex, multidimensional, and interacting biological, psychological, and social factors when treating patients, with an emphasis on relationships as central to healthcare delivery. Although participants did not specifically use the term ‘biopsychosocial approach’, they described the impact of clinician expressions that were aligned with and inverse to it (eg, expressions of listening, empathy, affirmation, engaging in open and honest conversations, and offering treatment options to the patient; see online supplemental material).

Archetypes and risk perception

Archetypes varied in how participants perceived the risk of harm from opioids, both for themselves and for others using opioids. Of the 40 participants, 11 (27.5%) reflected archetype 1 at some point in their lives (using opioids for chronic pain). These participants recognised there was a risk, but viewed that risk as acceptable because they needed the opioid to manage their pain and be able to engage in their daily activities. They described their risk as minimal because they were taking the medication as prescribed by their doctor for debilitating pain. They considered other patients whom they perceived as misusing or abusing opioids while not needing them for pain as the people at actual risk.

21 participants (52.5%) reflected archetype 2 at some point in their life. They recognised the risk opioids pose to themselves and others and wanted to avoid taking them chronically. Nine participants (22.5%) reflected archetype 3 at some point and 14 participants (35.0%) reflected archetype 4, with both groups recognising the risk opioids posed to themselves and others. Eight participants (20.0%) reflected archetype 5, and seven participants (35.0%) reflected archetype 6 at some point in their lives, with both groups perceiving opioids as risky and dangerous to themselves and others. They wanted to avoid any opioid exposure in the future to avoid risking becoming addicted again. They expressed most strongly that clinicians need to be straightforward with patients about the risk of opioids.

Having clinicians present information about opioids was just as important to participants as the tenor of the interaction. The most frequently mentioned information included risks and side effects, opioid safety and signs of misuse or withdrawal. Valence rankings for themes along with supportive quotes are provided in online supplemental material.

Discussion

The modified six archetype heuristic provides a pragmatic framework that may improve clinician communication and care activities for patients with or at high risk for OUD. The use of patient-preferred language and terminology by PCCs and practice staff can reduce the risk of patient treatment access limitations and recovery outcomes.³⁵ The results suggest Hooker et al’s archetypes²⁴ transferred well to this study sample and hinted at archetype differences in OUD-terminology preferences and opioid risk perceptions among participants. Of the four characteristics differentiating the original four archetypes, the only difference was that several participants in archetype 3 perceived opioids as necessary to manage pain.²⁴ Our results supported adding two additional archetypes representing participants in treatment or recovery, differentiated by whether they took MOUDs. Thus, three additional characteristics of ‘evidence of or confirmed OUD’, ‘in treatment/recovery’ and ‘taking an MOUD’ were added to the overarching framework to account for this difference. The adjustment conceptually aligns the archetype framework with the TTM, a leading evidence-based theoretical model described in figure 1 that has informed effective substance-use disorder care approaches such as motivational interviewing.^20–23

Figure 1. Conceptual alignment of proposed archetypes with the stages and processes of the transtheoretical model of health behaviour change. 20 21 A#, archetype # (e.g., A1, A2, A3…); MOUD, medication for opioid use disorder; OUD, opioid use disorder; TTM, transtheoretical model.

Transferability of Hooker et al’s four archetypes

Transferability in qualitative research refers to evidence a study’s findings apply to another sample based on population, setting or other shared characteristics.^{24 27 28}Individuals matching the four archetypes in both samples consistently viewed ‘addiction’ as stigmatising and less preferable to the term ‘dependence.’ The interstudy alignment was especially prominent among participants fitting the ‘long-term opioid use’ archetype, who commonly feared or experienced judgement from PCCs for their opioid usage. Second, participants in both studies shared opioid risk perceptions.^{24 36 37} Finally, participants in both studies, regardless of archetypes, wanted respectful, compassionate and open communication with their PCCs.

Several participants that best-fit archetype 3 perceived opioids as necessary to manage their pain, whereas Hooker et al’s framework reported this archetype did not perceive opioids as necessary to manage their pain.²⁴ The higher number of participants in our study may have revealed this difference while also assuming participants were appropriately provided non-opioid alternative to manage their pain. Our study also found evidence supporting expanding the framework to separate individuals who have taken action or are maintaining treatment for OUD from those who are contemplating or intending to change their relationship with opioids through treatment. Table 1 depicts the modified framework.

The six archetypes generated from our analysis aligned well with concepts from TTM. Juxtaposing TTM’s 6 stages and 11 processes of change with the 6 archetypes may reveal OUD-specific insights for clinician communication strategies not apparent in either independently, especially as it pertains to OUD-terminology preferences and opioid risk perceptions across the archetypes.

OUD-terminology preferences across patient archetypes

Terminology like ‘addict’ or ‘addiction’ was commonly viewed as stigmatising language that should be avoided in care settings.³⁵However, most study participants fitting the treatment/recovery archetypes held neutral or positive perceptions of these terms for patients in treatment/recovery. This result may reflect a reclamation of pejorative labels by many in the treatment/recovery community as an empowering act of honesty and humility.³⁸ It may be helpful for PCCs to be aware of the wide variation in the perception and impact of language and terminology among patients with or at risk of OUD. Our findings suggest the use of addiction terminology with a person in treatment or recovery can strengthen the therapeutic alliance in the maintenance stage of change, whereas a person with problematic opioid use who is open to treatment may be offended and perhaps revert to the precontemplation stage from the contemplation stage. Several participants in the study felt they were viewed and treated poorly because of their opioid use, which may have caused them to delay or avoid seeking necessary medical care. PCCs should seek to understand the patients’ language preferences and how they wish to be described,^{35 38} for which the archetypes heuristic could be helpful.

Adopting patients’ preferred terminology and how the patient defines it may contrast with how these terms are formally used and clinically defined. For instance, the formal clinical meaning of ‘addiction’ is an intense neurobiological urge to take a drug resulting in an individual’s loss of control even in the face of adverse consequences.^{39 40} Many participants associate ‘addiction’ with using opioids to get high in contrast to the lived experience of using opioids to avoid symptoms of withdrawal, which they describe as ‘dependence’. We argue the intended meaning of the clinician using the term is secondary to the patient’s understanding of it and the meaning assigned to it. Many study participants shared how the term ‘addiction’ evoked feelings of guilt or shame because it suggests a moral weakness to make a different choice. Other study participants identified as having or being high risk for OUD disliked ‘addiction’ because they resented being inappropriately grouped with people using opioids to get high.

The term ‘dependence’ was generally preferred among non-treatment/recovery study participants regardless of whether the latter term elicited feelings of shame or inappropriate judgement. However, study findings show the meaning of ‘dependence’ differed between study participants and the clinical literature. Study participants, especially those with chronic pain, commonly defined ‘dependence’ as reflecting how the drug allowed them to be active and functioning individuals. In contrast, the biophysical definition of ‘dependence’ is a process when a drug the body has become accustomed to is removed, resulting in withdrawal symptoms.^{39 40} Both definitions of ‘dependence’ seem to avoid the stigma associated with addiction’ and its negative impact on care and outcomes, so the differences in meaning between the colloquial and formal clinical definitions of ‘dependence’ have less pragmatic importance in treatment access and recovery outcomes. However, the lack of stigma may also open a door for both patient and clinician education to improve communication and understanding in opioid-related care within primary care settings.

Opioid-risk perceptions across patient archetypes

Patients’ perceptions of opioid risks can vary considerably, ranging from those who are unaware opioids have any risks to those who fear the associated risks of addiction and overdose so much they avoid opioids altogether for pain control.⁴¹ As universal approaches to risk communication can be problematic, the archetypes heuristic may be useful for PCCs and practice staff to tailor communication and care activities according to patients’ pain and treatment experiences.^{42 43} For instance, patients using opioids for chronic pain often recognised some risk for addiction and overdose that risk but viewed that risk as acceptable when faced with being unable to work, care for their families or engage in hobbies. Thus, PCCs should explicitly identify and appreciate these trade-offs in any discussion about opioid-related risks (ie, consciousness raising), problematic use (eg, environmental re-evaluation) and treatment (eg, self-re-evaluation). Patients who best fit the ‘in treatment/recovery’ and ‘openness to treatment’ archetypes more commonly held higher opioid risk perceptions and may therefore be receptive to straightforward discussions about the risk of opioids (eg, reinforcement management, counterconditioning, stimulus control).

Recommendations

The archetypes heuristic may help clinicians take a person-centred care approach by anticipating variances in language needs and/or preferences across patients with OUD (ie, sensitivities, potential harm and opportunities for empowerment, buy-in and trust building with trust strength-based language). Information contained in the heuristic can also provide PCCs clues for navigating discussions about the rationale, benefits and challenges in prescribing/using opioids, MOUDs, non-opioid drug therapy and non-pharmacological strategies. This should include what to do in emergencies when pain or withdrawal symptoms flare up.^{44 45} Opportunities exist to develop new educational resources or enhance existing resources using the insights from this work about best practices for PCCs and staff when interacting with and caring for patients with OUD or at risk for OUD.^{46 47}

Strengths and limitations

This study adopted a qualitative research assumption that the human experience comprises multiple subjective realities that are transferable with adequate description. It was also assumed that interview responses from patients were truthful and accurately reflected their experiences related to the study objectives without being returned to participants for comments and/or corrections.

The methodological design of the study included notable strengths and limitations. Strengths included a robust and systematic approach to collecting and analysing qualitative data collected from two separate health systems in geographically dispersed states. However, participants were limited to two healthcare systems that serve predominantly white, rural populations, which potentially limits generalisability of the findings. The findings from this study cannot be generalised to other population or settings. However, as the opioid epidemic has a high prevalence in white rural populations, these findings likely represent generalisable knowledge within that population and are important to address. Finally, it is important to acknowledge that the evidence supporting this heuristic does not account for all factors potentially affecting the quality and effectiveness of communication between PCCs/practice staff and each patient with OUD (eg, social determinants of health, comorbidities, family and friend support).^48–50

Future research is needed to validate the archetypes in larger and more diverse primary care populations as well as how to best integrate the framework into the OUD-CDS. Additional operational field testing of the OUD-CDS informed by our six-archetype model can be used to evaluate the framework’s effect on OUD-related communication quality and corresponding health outcomes.¹³

Conclusion

Discussions about opioids and OUD between patient and clinicians are important to care quality and outcomes.^14–17 The findings of this study support Hooker et al’s²⁴ archetype framework and identified two additional important types. The updated six-archetype framework may help clinicians and practice staff more effectively navigate conversations with patients diagnosed with or at high risk for OUD by considering how to discuss opioid risks and use opioid-related terminology preferred by the patient. Additional research into how the archetype framework connects to existing theoretical frameworks for facilitating behaviour change in care for individuals with substance abuse is needed.

The authors thank G Bart, A Bergdall, A Boelter, C Borgert-Spaniol, L Crain, S Dehmer, H Ekstrom, K Huntley, J McCormack, J Muegge, P O’Connor and J Sperl-Hillen for contributions to the clinical trial that informed this study. The authors also acknowledge utilisation of the AI language model tool Elicit to comprehensively search and identify literature connected to study findings.

Data availability statement

Data are available on reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The Institutional Review Boards (IRB) at all three systems reviewed and approved the trials (Essentia Health IRB#00001080, STUDY#EIRH-21-1880; Geisinger IRB#00008345, STUDY#2020-0436; HealthPartners Institute FWA#00000106, STUDY# A18-345), with oversight then ceded to the HealthPartners IRB.

Footnote

X @rxeric.wright

Contributors AWO: guarantor, conceptualisation, methodology, validation, investigation, resources, data curation, writing–original draft, writing–review and editing; AB: validation, investigation, data curation, writing–review and editing; CIA, LDT, HAH-B and CMG: validation, investigation, data curation, writing–review and editing, project administration; SAH: conceptualisation, writing–review and editing; LAF: investigation, data curation, writing–review and editing; project administration; RCR: conceptualisation, resources, writing–review and editing, funding acquisition; LIS: conceptualisation, writing–review and editing; EAW and IVH: resources, writing–review and editing; KMR: conceptualisation, methodology, investigation, resources, data curation, writing –o riginal draft, writing–review and editing, visualisation. The authors used the AI language model tool Elicit to comprehensively search and identify literature connected to the study findings.

Funding This work was supported by the National Institutes of Health (NIH) through the NIH Helping to End Addiction Long-term (HEAL) initiative under award number UG1da040316.

Disclaimer The content is solely the responsibility of the authors and does not necessarily represent the official views of NIH or the NIH HEAL initiative.

Competing interests None declared.

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Spencer M, Miniño A, Warner M. Drug Overdose Deaths in the United States, 2001–2021. 2023; 22. doi:10.15620/CDC:122556

² Centers for Disease Control and Prevention. U.S. Overdose deaths decrease in 2023, first time since 2018. national center for health statistics pressroom. 2024. Available: https://www.cdc.gov/nchs/pressroom/nchs_press_releases/2024/20240515.htm#:~:text=The

³ Council on Foreign Relations. Fentanyl and the U.S. Opioid Epidemic, Available: https://www.cfr.org/backgrounder/fentanyl-and-us-opioid-epidemic

⁴ Palumbo SA, Adamson KM, Krishnamurthy S, et al. Assessment of Probable Opioid Use Disorder Using Electronic Health Record Documentation. JAMA Netw Open 2020; 3: e2015909. doi:10.1001/jamanetworkopen.2020.15909

⁵ Dydyk A, Jain N, Gupta M. Opioid use disorder. In: StatPearls. StatPearls Publishing, 2023.

⁶ Center for Behavioral Health Statistics. Key Substance Use and Mental Health Indicators in the United States: Results from the 2021 National Survey on Drug Use and Health. 2022; EP22–07.

⁷ Centers for Disease Control and Prevention. CDC’s clinical practice guideline for prescribing opioids for pain. opioid prescribing resources. 2022. Available: https://www.cdc.gov/opioids/healthcare-professionals/prescribing/guideline/index.html

⁸ Austin EJ, Briggs ES, Ferro L, et al. Integrating Routine Screening for Opioid Use Disorder into Primary Care Settings: Experiences from a National Cohort of Clinics. J Gen Intern Med 2023; 38: 332–40. doi:10.1007/s11606-022-07675-2

⁹ Meunier PY, Raynaud C, Guimaraes E, et al. Barriers and Facilitators to the Use of Clinical Decision Support Systems in Primary Care: A Mixed-Methods Systematic Review. Ann Fam Med 2023; 21: 57–69. doi:10.1370/AFM.2908/-/DC1

¹⁰ Agency for Healthcare Research and Quality. Clinicial decision support. 2024 Available: https://www.ahrq.gov/cpi/about/otherwebsites/clinical-decision-support/index.html

¹¹ Rossom RC, Sperl-Hillen JM, O’Connor PJ, et al. A pilot study of the functionality and clinician acceptance of a clinical decision support tool to improve primary care of opioid use disorder. Addict Sci Clin Pract 2021; 16: 37. doi:10.1186/s13722-021-00245-7

¹² Bart GB, Saxon A, Fiellin DA, et al. Developing a clinical decision support for opioid use disorders: a NIDA center for the clinical trials network working group report. Addict Sci Clin Pract 2020; 15. doi:10.1186/s13722-020-0180-2

¹³ Rossom RC, Crain AL, O’Connor PJ, et al. Design of a pragmatic clinical trial to improve screening and treatment for opioid use disorder in primary care. Contemp Clin Trials 2023; 124: S1551-7144(22)00338-X. doi:10.1016/j.cct.2022.107012

¹⁴ Haug NA, Bielenberg J, Linder SH, et al. Assessment of provider attitudes toward #naloxone on Twitter. Subst Abus 2016; 37: 35–41. doi:10.1080/08897077.2015.1129390

¹⁵ Albrecht JS, Khokhar B, Pradel F, et al. Perceptions of patient provider agreements. J Pharm Health Serv Res 2015; 6: 139–44. doi:10.1111/jphs.12099

¹⁶ De Ruddere L, Craig KD. Understanding stigma and chronic pain: a-state-of-the-art review. Pain 2016; 157: 1607–10. doi:10.1097/j.pain.0000000000000512

¹⁷ Garpenhag L, Dahlman D. Perceived healthcare stigma among patients in opioid substitution treatment: a qualitative study. Subst Abuse Treat Prev Policy 2021; 16: 1–12. doi:10.1186/S13011-021-00417-3/METRICS

¹⁸ Danielson EC, Mazurenko O, Andraka-Christou BT, et al. An Analysis of Primary Care Clinician Communication About Risk, Benefits, and Goals Related to Chronic Opioid Therapy. MDM Policy Pract 2019; 4: 2381468319892572. doi:10.1177/2381468319892572

¹⁹ Mathis SM, Hagaman A, Hagemeier N, et al. Provider-patient communication about prescription drug abuse: A qualitative analysis of the perspective of prescribers. Subst Abus 2020; 41: 121–31. doi:10.1080/08897077.2019.1635956

²⁰ Prochaska JO, Velicer WF. The transtheoretical model of health behavior change. Am J Health Promot 1997; 12: 38–48. doi:10.4278/0890-1171-12.1.38

²¹ Marshall SJ, Biddle SJ. The transtheoretical model of behavior change: a meta-analysis of applications to physical activity and exercise. Ann Behav Med 2001; 23: 229–46. doi:10.1207/S15324796ABM2304_2

²² Prochaska JO, DiClemente CC, Norcross JC. In search of how people change. Applications to addictive behaviors. Am Psychol 1992; 47: 1102–14. doi:10.1037//0003-066x.47.9.1102

²³ Taş F, Seviğ EÜ, Güngörmüş Z. The effect of Transtheoretical Model based motivational interviewing on smokeless tobacco cessation in high school students. J Subst Use 2020; 25: 639–43. doi:10.1080/14659891.2020.1760372

²⁴ Hooker SA, Essien IJ, Borgert-Spaniol CM, et al. Patient Perspectives on Opioid Risk Discussions in Primary Care. J Patient Cent Res Rev 2022; 9: 253–62. doi:10.17294/2330-0698.1941

²⁵ Benyamin R, Trescot AM, Datta S, et al. Opioid complications and side effects. Pain Physician 2008; 11: S105–20.

²⁶ Campbell S, Greenwood M, Prior S, et al. Purposive sampling: complex or simple? Research case examples. J Res Nurs 2020; 25: 652–61. doi:10.1177/1744987120927206

²⁷ Guba EG. Criteria for assessing the trustworthiness of naturalistic inquiries. ECTJ 1981; 29: 75. doi:10.1007/BF02766777

²⁸ Krefting L. Rigor in qualitative research: the assessment of trustworthiness. Am J Occup Ther 1991; 45: 214–22. doi:10.5014/ajot.45.3.214

²⁹ Saunders B, Sim J, Kingstone T, et al. Saturation in qualitative research: exploring its conceptualization and operationalization. Qual Quant 2018; 52: 1893–907. doi:10.1007/s11135-017-0574-8

³⁰ TranscribeMe Inc. TranscribeMe! 2023.

³¹ Watkins D. Rapid and Rigorous Qualitative Data Analysis: The “RADaR” Technique for Applied Research. Int J Qual Methods 2017; 16: 1–9. doi:10.1177/1609406917712131

³² Forbes CM, Bonnet K, Bryant T, et al. Decision-making and improvements in health-related quality of life in patients with kidney stones: comparing surgery versus observation using a mixed methods analysis. Urolithiasis 2022; 50: 567–76. doi:10.1007/s00240-022-01339-w

³³ Olson AW, Stratton TP, Isetts BJ, et al. Seeing the Elephant: A Systematic Scoping Review and Comparison of Patient-Centeredness Conceptualizations from Three Seminal Perspectives. J Multidiscip Healthc 2021; 14: 973–86. doi:10.2147/JMDH.S299765

³⁴ Mead N, Bower P. Patient-centredness: a conceptual framework and review of the empirical literature. Soc Sci Med 2000; 51: 1087–110. doi:10.1016/s0277-9536(00)00098-8

³⁵ National Institute on Drug Abuse. Words matter: preferred language for talking about addiction. Advanciing addiction science. 2021. Available: https://nida.nih.gov/research-topics/addiction-science/words-matter-preferred-language-talking-about-addiction#references

³⁶ Behar E, Rowe C, Santos GM, et al. Primary Care Patient Experience with Naloxone Prescription. Ann Fam Med 2016; 14: 431–6. doi:10.1370/afm.1972

³⁷ Rowe C, Santos G-M, Behar E, et al. Correlates of overdose risk perception among illicit opioid users. Drug Alcohol Depend 2016; 159: 234–9. doi:10.1016/j.drugalcdep.2015.12.018

³⁸ Ashford RD, Brown AM, Curtis B. The Language of Substance Use and Recovery: Novel Use of the Go/No-Go Association Task to Measure Implicit Bias. Health Commun 2019; 34: 1296–302. doi:10.1080/10410236.2018.1481709

³⁹ Szalavitz M, Rigg KK, Wakeman SE. Drug dependence is not addiction-and it matters. Ann Med 2021; 53: 1989–92. doi:10.1080/07853890.2021.1995623

⁴⁰ O’Brien CP, Volkow N, Li T-K. What’s in a word? Addiction versus dependence in DSM-V. Am J Psychiatry 2006; 163: 764–5. doi:10.1176/ajp.2006.163.5.764

⁴¹ Conrardy M, Lank P, Cameron KA, et al. Emergency Department Patient Perspectives on the Risk of Addiction to Prescription Opioids. Pain Med 2016; 17: 114–21. doi:10.1111/pme.12862

⁴² Daniulaityte R, Falck R, Carlson RG. “I’m not afraid of those ones just 'cause they’ve been prescribed”: perceptions of risk among illicit users of pharmaceutical opioids. Int J Drug Policy 2012; 23: 374–84. doi:10.1016/j.drugpo.2012.01.012

⁴³ McDonnell E, Harmon D. A systematic review of chronic pain patients’ perceptions of prescription opioids. M Perit 2020; 4: AB112. doi:10.21037/map.2020.AB112

⁴⁴ Pandey M, Marwah R, McLean M, et al. Patient perspectives from the multi-disciplinary chronic pain clinic: a qualitative study. Pain Manag 2022; 12: 383–96. doi:10.2217/pmt-2021-0072

⁴⁵ Pandey M, Paluck E, Maierhoffer S, et al. Lessons learned from the chronic pain clinic: a qualitative study of the perspectives of healthcare providers. Pain Manag 2023; 13: 457–71. doi:10.2217/pmt-2023-0048

⁴⁶ Klusaritz H, Bilger A, Paterson E, et al. Impact of Stigma on Clinician Training for Opioid Use Disorder Care: A Qualitative Study in a Primary Care Learning Collaborative. Ann Fam Med 2023; 21: S31–8. doi:10.1370/AFM.2920/-/DC1

⁴⁷ Zittleman L, Curcija K, Nease DE, et al. Increasing Capacity for Treatment of Opioid Use Disorder in Rural Primary Care Practices. Ann Fam Med 2022; 20: 18–23. doi:10.1370/afm.2757

⁴⁸ Common comorbidities with substance use disorders research. 2020. Available: https://www.ncbi.nlm.nih.gov/books/NBK571451/

⁴⁹ Pettersen H, Landheim A, Skeie I, et al. How Social Relationships Influence Substance Use Disorder Recovery: A Collaborative Narrative Study. Subst Abuse 2019; 13: 1178221819833379. doi:10.1177/1178221819833379

⁵⁰ Rottnek F, Bello-Kottenstette JK. The Impact of Social and Structural Determinants on Depression, Prescription Opioid Use, Opioid Misuse, and Opioid Use Disorder. Pain, the Opioid Epidemic, and Depression 2024; 117–37. doi:10.1093/9780197675250.003.0007