Toward safer opioid prescribing: effects of the TOWER intervention on HIV care providers

Abstract

Rates of opioid use disorder and associated deaths remain alarmingly high. Measures to address the epidemic have included reductions in opioid prescribing, in part guided by the Centers for Disease Control Opioid Prescribing Guideline (CDCG). While reductions in over-prescribing have occurred, these measures have also resulted in decreased access and adverse outcomes for some stable opioid-treated chronic pain patients. The TOWard SafER Opioid Prescribing (TOWER) intervention was designed to support HIV primary care providers in use of the CDCG and in decision-making and patient–provider communication regarding safe opioid prescribing. Eleven HIV primary care providers and 40 of their patients were randomized into intervention and control groups. Transcripts from 21 patient visits were analyzed, focusing on opioid and pain-related communications. Findings from this research indicate greater alignment with the CDCG among visits carried out with providers in the TOWER intervention group. However, control group visits were notably consistent with guideline recommendations in several key areas. Differences observed between the intervention and control group visits demonstrate intervention strengths, as well as areas where additional work needs to be done to ensure prescribing and communication consistent with the CDCG.

Introduction

Opioid overdose deaths remain at epidemic levels, with over 46,000 lives lost in 2018 (Centers for Disease Control and Prevention, 2018; National Institute on Drug Abuse, 2020). Currently, over two million Americans are living with an opioid use disorder. Prescription opioids were a critical driver of the early U.S. opioid epidemic (Jones, 2013; Lankenau et al., 2012), however, prescribing rates have been steadily declining since 2012, with a more rapid rate of decline since the publication of the Centers for Disease Control and Prevention Opioid Prescribing Guideline (CDCG) in 2016 (Centers for Disease Control and Prevention, 2020; Dowell et al., 2016). Although a reduction in opioid prescribing was a necessary correction, it has also resulted in decreased access and adverse outcomes for many stable opioid-treated chronic pain patients (Comerci et al., 2018; Darnall et al., 2018; Samet & Kertesz, 2018; Weeks, 2016), arguing for interventions that support a more individualized and collaborative patient-centered approach.

Some of the CDCG recommendations, such as incorporation of non-pharmacologic pain treatments, are relatively specific and straightforward to implement. However, the CDCG also describes the importance of assessment of benefit, harm, and risk of harm from prescription opioids, which is a nuanced and time-consuming process that presents several challenges (Navis et al., 2019). Due to this complexity, interventions to improve opioid prescribing practices described in prior research have largely relied on ancillary staff (Liebschutz et al., 2017; Quanbeck et al., 2018; Witt et al., 2018) or customized electronic health record (EHR)-embedded decision support. While such resources have been shown to be beneficial in supporting specific provider behaviors such as reducing the average morphine milligram equivalents (MME) prescribed, they may not be widely available (Anderson et al., 2015). Moreover, it is uncertain how such interventions perform with complex patients or historically marginalized populations, such as people living with HIV.

In order to address these limitations, we created the TOWard SafER Opioid Prescribing (TOWER) intervention, which was designed to support HIV primary care providers in CDCG adherence. Development of TOWER, which was based on the Information, Motivation and Behavioral Skills (IMB) Model, and used an iterative process of patient and provider input, has been described previously, along with the design of the pilot clinical trial assessing its feasibility (Robinson-Papp et al., 2019). Briefly, TOWER is a three-part provider-focused intervention that consists of: (1) a provider training, including patient-recommended communication strategies for prescription opioid management (Scherer et al., 2019); (2) access to patient-reported opioid risk-benefit data via a novel opioid management app (OM-App); and (3) an opioid management note template (OM-Note), which guides the provider through a CDCG-adherent opioid management visit.

The design of the randomized, controlled pilot clinical trial to test TOWER’s feasibility and optimize its content has been described elsewhere (Robinson-Papp et al., 2019); quantitative feasibility results are forthcoming. Here, we describe findings from the trial’s qualitative component, which used audiotaped patient–provider clinic visits to examine communication patterns relevant to the CDCG topic areas.

Methods

This paper focuses on one of the final components of the TOWER study: a comparison of visit content among providers trained on TOWER (intervention group) and providers providing usual care (control group). The main topic areas intervention providers were asked to address in their interactions with people living with HIV to whom they prescribed opioids for chronic pain are outlined in Table 1 below. Intervention group providers received a training that lasted approximately 75 min and focused on the content of the CDCG, use of the OM-App and OM-note, and communication strategies to help them move through the topics outlined in Table 1. They were instructed to use the materials during regularly scheduled visits with enrolled patients. Control group providers were instructed to follow their usual care practices.

Table 1.

Topic areas.

1. Acknowledge this visit will be different than usual (communication strategy)

2. Elicit patient concerns using open-ended questions (communication strategy)

3. Discuss alternative treatments (non-opioid pharmacologic and non-pharmacologic)

4. Discuss pain severity (intensity and impact on function)

5. Discuss identified treatment goals and assess progress

6. Assess benefit of current opioid pain treatment

7. Discuss adherence to prescribed opioid

8. Evaluate harms/risk factors for harm of prescribed opioid

9. Discuss opioid-related harms (opioid use disorder, overdose)

10. Discuss monitoring: PDMP and urine drug test

11. Assess morphine equivalency

12. Discuss opioid refill/increase/taper

13. Discuss risk mitigation: Naloxone & plan for using opioids safely

14. Schedule follow-up visit

Eleven HIV Primary Care Providers (PCPs) and 40 of their respective patients participated in the TOWER study; 6 PCPs were randomized to intervention and 5 to control. All patients were living with HIV and chronic pain and were taking opioids prescribed and monitored by one of the study providers. Providers were randomized into an intervention and control group by research staff. Participating providers and patients consented to have a visit observed by a member of the research team, audio recorded, and professionally transcribed. Findings presented here are based on transcripts from 21 patient visits with providers: 10 intervention group visits and 11 control group visits. Although attempts were made to audiotape one visit with each study participant this was not always possible due to scheduling difficulties (including the beginning of the COVID-19 pandemic).

Transcripts were managed, coded and analyzed using NVivo 12 (Melbourne, Australia). Coding was done by multiple members of the research team and discussed regularly to ensure consistency. Inductive and deductive processes were used to develop a coding scheme and to identify themes most consistent with the project aims, while also allowing for concepts to emerge from the data themselves. The coding scheme covered all the main topic areas included in the TOWER intervention.

Analysis focused on how (and how often) intervention group providers covered intervention topic areas in patient encounters; barriers and facilitators to implementation; understanding what constitutes usual care among control group visits; and patient–provider communication about opioids more generally. As part of the analysis, we indicated for all of the transcripts whether each of the main TOWER topic areas was discussed or not.

Enrolled patients received $50 for all in-person visits with a member of the research team and $25 per month if they completed at least 80% of the online OM-App questions. Participating PCPs received $750 for their participation in the study. The research was approved by the Institutional Review Boards of the Icahn School of Medicine at Mount Sinai and The New York Academy of Medicine. All provider and patient participants gave written informed consent.

Findings

As shown in Table 2 below, all TOWER topics were covered in multiple study visits. Topics covered in 70% or more of the intervention visits included a range of intervention themes, namely use of non-opioid treatment approaches (n = 8), discussion of pain severity (n = 8) and treatment goals (n = 7), adherence to the prescribed opioid (n = 8), the impact of current opioid pain treatment (n = 9), evaluation of risks and harms related to opioid use (n = 8), and monitoring tools such as use of a Prescription Drug Monitoring Program (PDMP) and urine toxicology testing (n = 8).

Table 2.

Coverage of intervention topics during clinical visits.

	Frequency
	Intervention (N = 10)		Control (N = 11)
Topic areas	n	%	n	%
Acknowledge this visit will be different than usual (communication strategy)	4	40%	n/a
Elicit patient concerns using open-ended questions (communication strategy)	3	30%	6	55%
Discuss alternative treatments (non-opioid pharmacologic and non-pharmacologic)	8	80%	9	82%
Discuss pain severity (intensity and impact on function)	8	80%	9	82%
Discuss identified treatment goals and assess progress	7	70%	1	9%
Assess benefit of current opioid pain treatment	9	90%	5	45%
Discuss adherence to prescribed opioid	8	80%	0	0%
Evaluate harms/risk factors for harm of prescribed opioid	8	80%	1	9%
Discuss opioid-related harms (opioid use disorder, overdose)	6	60%	0	0%
Discuss monitoring: PDMP and urine drug test	8	80%	2	18%
Assess morphine equivalency	4	40%	0	0%
Discuss opioid refill/increase/taper	6	60%	8	73%
Discuss risk mitigation: Naloxone & plan for using opioids safely	5	50%	1	9%
Schedule follow-up visit	5	50%	9	82%

Control group visits consistently covered fewer of the intervention topics than the intervention group. Topics covered in 70% or more control group visits included discussion of alternative treatments (n = 9), pain severity (n = 9), discussion of opioid refill/increase/taper (n = 8) and scheduling a follow-up visit (n = 9).

Topic areas with the greatest discrepancies between intervention and control included discussions of treatment goals and assessment of progress (7 intervention visits compared to 1 control visit), adherence to prescribed opioids (8 intervention visits compared to 0 control visits), evaluation of harms/risk factors for harm of prescribed opioids (8 intervention visits compared to 1 control visit) and discussion of opioid-related harms such as opioid use disorder and overdose (6 intervention visits compared to 0 control visits).

There was substantial variability in how the topics were addressed during the visits, with some providers stating information seemingly without opportunity for discussion, and others encouraging more extensive conversation, including questions regarding progress or other changes over time. When asking about patient risk factors for harm, however, providers often used closed-ended questions without exploring patient responses. Examples from the transcripts, categorized by intervention topic area, are provided in Table 3.

Table 3.

Excerpts from patient visit transcripts.

Alternative treatments
	Provider: So, you do exercises? Patient: Yes. Provider: What kind of exercises? Patient: Squats. Leg lifts. Provider: Who instructed you on those things? Patient: When I went through physical therapy, you know how they give you little paper pamphlets to do exercise after the therapy is over that you can start doing them? So, I memorized them so well.
Pain severity
	Provider: And you’re still taking the patches, but – oh, so your pain is at a six, seven? Patient: Yeah. Provider: Would you say that’s less for you than before or about the same? Patient: I don’t know, I say that it’s about the same. I’ve been doing a lot more walking, but the pain is the same. Well, no, it’s been hurting a lot at night.
Progress toward treatment goals		Provider: Okay. All right. So, tell me in your own words, why I’m giving you – what the goal is for the medicine, the opiates I’m giving you? Like what do you think it should be doing? Patient: It should be taking away, you know, modifying the pain in my right knee, so my leg wouldn’t give out or get stiff.
Benefit of current opioid use	Provider: Okay. Now, do any of these medications that I’m giving you – the morphine and the Vicodin, do they control your pain adequately? Patient: Yes. They do. Provider: Okay. Do you need more? Patient: No. I don’t. No. I just got them the other day. And I need them, thank god, because they didn’t give me anything for the pain after the surgery.
Adherence to prescribed opioid	Provider: You’re doing the six pills a day, right, as much as supposed to? Patient: Yeah, mm-hmm.
Assessing risks and harms	Provider: So, you’re not using any other drugs? Patient: No. Provider: No recent overdose? No. Sedation and intoxication? No. Benefits of opioids? Yes, I think it seems to be helping you a little bit.
Opioid-related harms (opioid use disorder, overdose)	Provider: Okay. Notes on Ambien. I’m just gonna write down. Do you have any previous history of substance abuse? Patient: No. Provider: IV drug use? Patient: No. Provider: Nothing like that? Patient: No. Provider: Okay. Any family history of substance abuse– Patient: Older brother. … Provider: Do you ever lose control? Are you compulsive in what you do? Do you use a lot of your Vicodin or morphine –? Patient: No. I don’t hardly use it.
Monitoring: PDMP and urine drug test		Provider: All right. You’re going to do a urine test, also. Okay? Patient: I’ll try to. I went before I went home but if I drink some water, that will probably make me go. Provider: Just to make sure you have the medicine in you.
Assess morphine equivalency	Provider: And you are on Vicodin and morphine? Patient: Mm-hmm. Provider: Correct? Patient: Yeah. Provider: And the opioid medications are Vicodin and morphine. And you’re morphine equivalent – I have the app. So, your morphine is 30 q 8, so that’s 90 mg.
Opioid refill/increase/taper	Provider: I don’t know that I feel comfortable increasing your pain medication. I might have you see someone here to evaluate you to figure out whether you’re on the best pain regimen.
Risk mitigation		Provider: I wanna make sure … yeah, just – this [Naloxone prescription] is just for protection. To benefit you … I’m gonna give you four sprays. Okay? Patient: Mm-hmm. Provider: For each refill. It’s just to protect you. Repeat every – what’s it say? It says, “Repeat every two to three minutes.” Repeat every two to three minutes until patient is responsive. You live with someone at home? Or you’re just by yourself? Patient: Just by myself. Patient: Just by myself.

Discussion

Findings from this research demonstrate greater alignment with CDCG among visits carried out with providers in the TOWER intervention group than in the control group. However, control group visits were also generally consistent with guideline recommendations in several key areas.

Variability in coverage of CDCG topics observed within the intervention group likely suggests which aspects of the TOWER intervention may be easiest to incorporate into healthcare practice. Specifically, intervention providers consistently addressed alternative treatments, pain severity, the impact of current opioid pain treatment, patient adherence to prescribed opioids, evaluation of harms/risk factors for harm and prescription monitoring (i.e., PDMP and/or urine drug testing). Other than discussing pain severity and alternative treatments – both of which were addressed with similar regularity in control visits – the topics outlined here represent topics or practices successfully incorporated into the intervention visits.

Differences between intervention and control groups most consistently related to discussion of treatment goals, opioid adherence, evaluation of risks and harms – including possible opioid use disorder or overdose – and mitigation of opioid risk. The consistent appearance of these topics in intervention encounters compared to those in the control group provides promising evidence that interventions such as TOWER may successfully promote discussion consistent with the CDCG regarding opioid prescribing.

Notably, control group providers assessed pain severity and discussed alternative pain therapies in roughly equal measure to intervention group providers, suggesting they may already be incorporated into chronic pain management practice. Research related to opioid prescribing in other chronic pain management settings support this supposition (Kopp et al., 2020).

Although this study focused on the topics covered rather than the specific language used, we do note that providers often used brief, closed-ended statements when discussing patient risks and harms. Open-ended questions that allow the patient to fully communicate their experience and perspectives are an essential component of high-quality health care (Boyle et al., 2005). The short nature of these exchanges may point to the difficulty of incorporating an increased number of mandatory topics into provider–patient encounters, which often occur with limited time for lengthy discussion (Dysart & Tomlin, 2002). This may also indicate that when addressing difficult topics with their patients, providers default to discussing them in a perfunctory manner.

Several limitations to this work exist. First, this analysis is based on transcripts of in-person patient–provider interactions. We did not have access to background information that may have proved useful, such as: (1) the quality of the patient–provider relationship, including knowledge gained from one another in prior encounters; (2) body language, tone of voice, and movement, including facial expressions and physical references to the provider’s computer screen or the patient’s body and; (3) actions taken by the provider prior to, or following the encounter, such as use of PDMP or calculation of morphine equivalencies. Second, a member of the research team was present during the patient–provider visits, likely impacting provider and/or patient behavior (McCambridge et al., 2014) that may have increased observed fidelity to the intervention. Additionally, the presence of a researcher in the room allowed providers to rely on that person for support using the OM-Note. Alternatively, the researcher might have made the patient or provider feel less comfortable talking openly, particularly about opioid use, which carries both stigma and legal implications. Third, the sample size is small and, as a result, interpretation of the findings is limited. Finally, the amount of time between provider training and the recorded encounter varied, perhaps impacting the level of fidelity to and familiarity with the intervention across TOWER providers. Despite these limitations, findings demonstrate a positive impact of the intervention on provider adherence to the CDCG.

Results additionally highlight areas where intervention implementation was most successful and where greatest differences between intervention and control group visits exist. These variances may be used to refine the TOWER intervention as well as identify areas of priority for interventions that seek to promote provider adherence to the CDCG. Future interventions should also consider focusing on areas of greatest difference between the groups in order to maximize resources and increase the potential to impact how providers approach opioid prescribing and conduct related patient visits. These include discussion of treatment goals, assessment of benefit from current opioid treatment, evaluation of harms and risks, and use of monitoring tools. Finally, future iterations of the TOWER intervention may need to be adapted to include additional support related to patient–provider communication styles and to anticipate time-related challenges that may hamper provider ability to elicit patient input for opioid decision-making.