Study protocol for the targeting effective analgesia in clinics for HIV (TEACH) study – a cluster randomized controlled trial and parallel cohort to increase guideline concordant care for long-term opioid therapy among people living with HIV

Abstract

Background: People living with HIV (PLWH) frequently experience chronic pain and receive long-term opioid therapy (LTOT). Adherence to opioid prescribing guidelines among their providers is suboptimal.

Objective: This paper describes the protocol of a cluster randomized trial, targeting effective analgesia in clinics for HIV (TEACH), which tested a collaborative care intervention to increase guideline-concordant care for LTOT among PLWH.

Methods: HIV physicians and advanced practice providers (n = 41) were recruited from September 2015 to December 2016 from two HIV clinics in Boston and Atlanta. Patients receiving LTOT from participating providers were enrolled through a waiver of informed consent (n = 187). After baseline assessment, providers were randomized to the control group or the year-long TEACH intervention involving: (1) a nurse care manager and electronic registry to assist with patient management; (2) opioid education and academic detailing; and (3) facilitated access to addiction specialists. Randomization was stratified by site and LTOT patient volume. Primary outcomes (≥2 urine drug tests, early refills, provider satisfaction) were collected at 12 months. In parallel, PLWH receiving LTOT (n = 170) were recruited into a longitudinal cohort at both clinics and underwent baseline and 12-month assessments. Secondary outcomes were obtained through patient self-report among participants enrolled in both the cohort and the RCT (n = 117).

Conclusions: TEACH will report the effects of an intervention on opioid prescribing for chronic pain on both provider and patient-level outcomes. The results may inform delivery of care for PLWH on LTOT for chronic pain at a time when opioid practices are being questioned in the US.

Trial registration: ClinicalTrials.gov identifier: NCT02564341.

Trial registration: ClinicalTrials.gov identifier: NCT02525731.

Keywords:

• HIV
• chronic pain
• substance use disorders
• long-term opioid therapy
• prescription opioid misuse
• pain management
• addiction

Introduction

As overdose deaths in the United States have continued to rise, national attention on the opioid epidemic has led to intense scrutiny of the use of prescription opioids for chronic pain.^1–5 Prescription opioids have been identified as one of the chief contributors to the epidemic,⁶ and thus, the risk-benefit ratio must be carefully considered when treating chronic pain with long-term opioid therapy (LTOT).^7,8

People living with HIV (PLWH) sit at this junction, representing a high-need group in terms of their painful comorbid conditions, and a high-risk group in terms of potential adverse outcomes related to opioids due to intersecting risk factors such as substance use.⁹ PLWH have higher rates of chronic pain than the general population, with estimates ranging from 39% to 85%,^10–25 and prescription opioid rates as high as 65%.^26–29 Further, these patients receive opioids at higher doses than individuals without HIV, and are more likely to be co-prescribed benzodiazepines or other sedating medications.^27–32 Concomitant with the rise of prescription opioid misuse in the general population, the nonmedical use of these medications is increasingly common among PLWH.^33–37 Our group recently found that among PLWH on LTOT, 43% had a high current opioid misuse measure (COMM) score, suggestive of potential opioid misuse.^38,39 This is presumably related to the known co-morbidity between HIV and opioid use disorder.^40,41 At the same time, pain is frequently associated with substance use,^42–46 with some studies even suggesting that individuals self-medicate.^47,48 Chronic pain not treated with LTOT among PLWH has been associated with increased odds of suboptimal retention in care.⁴⁹ In contrast, LTOT has been associated with higher rates of virologic suppression, pointing to the potential HIV-related benefits that can occur in the context of prescribing opioids for pain.⁴⁹

Given the frequent coexistence of chronic pain and substance use risk in PLWH,^50,51 appropriate management of pain with LTOT is complicated.

HIV providers are faced with the complex and often daunting task of prescribing opioids for pain. National guidelines for prescribing LTOT exist, including those specifically for PLWH, and these guidelines can help providers make informed decisions to mitigate risks.^7,52 However, adopting LTOT guidelines is challenging for HIV providers.⁵³ They report low levels of satisfaction for pain management aspects of clinical care, as well as low confidence in their abilities to treat chronic pain.^54,55 Few physicians follow standard protocol regarding assessment and may lack resources to closely monitor patients,^55–60 resulting in wide variation in opioid prescribing patterns, similar to primary care.⁶¹

Although an urgent need exists for interventions that can improve HIV providers’ quality of care regarding opioid pain prescribing, limited studies of such interventions have been conducted to date. None, to our knowledge, focus on opioid prescribing for PLWH.^62–67 In order to address this need, we built upon the transforming opioid prescribing in primary care (TOPCARE) intervention for use in PLWH.^68,69 Liebschutz et al.^68,69 found that TOPCARE improved adherence to guideline-recommended monitoring among primary care patients at a safety net hospital and affiliated community health centers. However, patient-reported outcomes were not assessed. Using TOPCARE as a model, we developed the “Targeting Effective Analgesia in Clinics for HIV” (TEACH) intervention, geared toward HIV providers, and established an observational cohort of PLWH receiving LTOT to assess provider and patient outcomes related to LTOT. This paper describes the study design and implementation.

Methods

Study objective and design

We conducted a cluster-randomized clinical trial of the TEACH collaborative care intervention at two clinics (Boston Medical Center [BMC], Boston, MA and Emory University/Grady Hospital, Atlanta, GA). In parallel, PLWH receiving LTOT (n = 170) were recruited into a longitudinal cohort at both clinics and assessed at baseline and 12 months in order to collect outcomes not available in medical records. Together, the RCT and cohort enabled assessment of the following aims: Aim 1: assess HIV providers’ adherence to LTOT guidelines^8,70, as measured by occurrence of ≥2 urine drug tests (UDTs) at 12 months (primary); Aim 2: assess patient-level outcomes, as measured by percent of patients with any early refills at 12 months (primary); Aim 3: assess HIV providers’ satisfaction with prescribing LTOT (primary), as measured by mean satisfaction at 12 months; and Aim 4: assess HIV disease status among patients, as measured by undetectable viral load at 12 months (exploratory) and CD4 count (exploratory). Secondary RCT outcomes were obtained via provider self-report and patient self-report among patient participants enrolled in both the RCT and the cohort (Table 1). The RCT was implemented from September 2015 to December 2017. The patient cohort was recruited and assessed between July 2015 and February 2018. The overlap of the RCT and cohort study groups is depicted in Figure 1. The TEACH study was approved by the institutional review boards at Boston University Medical Campus, Emory University and the Grady [Health System] Research Oversight Committee.

Figure 1 Venn diagram describing overlap between targeting effective analgesia in clinics for HIV (TEACH) RCT patient participants and observational cohort of patient participants.

Table 1 TEACH study outcomes

Aim	Primary outcome	Secondary outcomes
1. Provider adherence to LTOT guidelines	≥2 UDT between baseline and 12 months (EMR data extraction)	1. ≥3 primary care visits in ID clinic (EMR data extraction)
		2. Proportion of patients with discontinuation of opioid prescriptions (EMR data extraction)
		3. Opioid treatment agreements (y/n) (EMR data extraction)
		4. % providers consulting Prescription Monitoring Program (provider self-report)
2. Patient level outcomes	% patients with ≥1 early refills at 12 months (EMR data extraction)	1. # early refills at 12 months (EMR data extraction)
		2. Aberrant use (patient cohort self-report)
		3. # patients with ED visits seeking opioids (EMR data extraction)
		4. Pain severity/interference (patient cohort self-report)
		5. Addiction severity (patient cohort self-report)
3. Provider and patient satisfaction with care	Provider satisfaction score (provider self-report)	1. Provider confidence scores (provider self-report)
		2. Patient LTOT satisfaction (patient cohort self-report)
		3. Patient trust in provider (patient cohort self-report)
4. HIV disease status		1. Undetectable viral load (exploratory) (EMR data extraction)
		2. CD4 count (exploratory) (EMR data extraction)

Cluster randomized controlled trial

Participants and recruitment

HIV physicians and advanced practice providers were recruited from September 2015 to December 2016 from two safety-net, hospital-based HIV clinics in Boston and Atlanta. Potential provider participants and their patients were identified by queries of the electronic medical record by the clinical data warehouse (CDW) at BMC and the Center For AIDS Research (CFAR) team at Emory University. Provider participant inclusion criteria included: (1) being an HIV physician or advanced practice provider (i.e. nurse practitioner or physician assistant) at the medical centers’ HIV clinics; and (2) being the main provider for ≥1 individual living with HIV who is receiving LTOT (defined as having received ≥3 opioid prescriptions ≥21 days apart within a 6-month period in the prior year). While tramadol is a schedule IV medication, reflecting a perception of less potential for abuse, it was included as a qualifying opioid for LTOT as it is an opioid. Provider participant exclusion criteria included: (1) being an investigator in this study; and (2) planning to leave the clinic <9 months from screening. In total, 41 providers were identified and all enrolled in the study. Patients receiving LTOT from physician participants (n = 187) were enrolled through a waiver of informed consent. The rationale for the waiver of informed consent was twofold. Because the trial was randomized at the level of the provider, it would have been logistically not possible for a provider who was receiving education, counseling, and access to a nurse care manager to selectively apply this knowledge and expertise to only certain patients. In addition, there was little risk to patients. These patient participants met the following eligibility criteria: (1) ≥18 years of age; (2) diagnosis of HIV infection from the EMR by ICD-9 codes or lab tests; (3) having received ≥3 opioid prescriptions ≥21 days apart within a 6-month period in the prior year from the study clinic site; and (4) having attended ≥1 visit to the enrollment sites within the prior 18 months. All patients who met the eligibility criteria for the clinical trial were allocated to the same treatment group (i.e. TEACH vs. control) as their provider. Study staff and clinician investigators reviewed the queried lists to confirm that those included were eligible for the study and correctly matched between patient and provider. The lists from each site were generated a second time near the end of the recruitment period to maximize the inclusion of all eligible providers and patients. Of the 187 participants enrolled into the RCT, 117 were simultaneously enrolled in the cohort (described below and Figure 1).

Research staff worked with clinic leadership to identify a regularly scheduled clinic meeting at which to present the study and enroll interested providers. Providers were informed ahead of the meeting that there would be a presentation about a new research study. The research teams then introduced the study at the regularly scheduled clinic meeting and invited eligible providers to enroll. It was reinforced that they could choose not to participate or discontinue participation in the study at any time, for any reason, with no penalty. Study staff followed up individually with providers who were unable to attend the clinic meetings to enroll them into the study. Provider enrollment is outlined in Figure 2.

Figure 2 Study flow diagram describing recruitment of HIV providers and their patients into the targeting effective analgesia in clinics for HIV (TEACH) RCT.

Research assessments

Providers completed 15-minute paper baseline and 12-month follow-up assessments, which included: demographics, training and practice characteristics, substance use among PLWH, practices for assessing and treating pain, and practices for managing use of prescribed opioids.^68,71,72 At follow-up, providers randomized to the intervention group also completed a brief evaluation assessment of the intervention. To ensure completion and minimize assessment bias, research staff were present to proctor providers as they completed surveys. Providers were compensated with a $100 gift card upon completion of each assessment. The paper assessment forms were double-entered into REDCap by research staff and any data entry discrepancies were resolved.⁷³

Randomization

After completing a baseline assessment, providers were randomized to either the control group or the year-long TEACH intervention. Randomization was stratified by site and patient volume (1–2, 3–6, 7–11, and 12+ patients). To ensure balance with respect to the number of providers in each group, the permuted blocks strategy was used with blocks of 2. Due to the nature of the intervention, neither provider participants nor the study team members were blinded to provider group assignment.

Control condition

At the clinic meeting when the study was introduced, all providers were given an informational brochure summarizing guidelines for LTOT and listing a web resource with electronic tools (Appendix).^8,68 Those who were randomized to the control group did not receive an intervention beyond the information about the study and the corresponding brochure. The control group did not have access to the support of the TEACH intervention.

Intervention

The 12-month TEACH intervention consisted of: (1) a nurse care manager and electronic registry to assist with patient management; (2) opioid education and academic detailing; and (3) facilitated access to addiction specialists. TEACH was grounded in the Chronic Care Model (CCM) for chronic disease management,^64,65 and based on the TOPCARE intervention and office-based addiction treatment (OBAT) model,^68,74 both based in primary care clinics. The CCM model seeks to improve patient outcomes through integrative systems changes, including organizational support, clinical information systems, delivery system redesign, decision support, self-management support and community resources. The CCM has been found effective for the care of a number of chronic diseases (e.g. tobacco use, diabetes, and depression),^75–77 and has served as a theoretical basis for a collaborative care intervention for a program to treat opioid-dependent patients with buprenorphine at BMC and community health clinics.^74,78 Central to this model is the use of a nurse care manager to provide support to providers. TOPCARE was a cluster-randomized clinical trial of nurse care management, an electronic registry, a single academic detailing session, and electronic decision support tools conducted in four safety-net primary care practices. TEACH was modeled after TOPCARE, with the following adaptations: this was conducted in an HIV clinic compared to a primary care clinic; TOPCARE lacked the ability to obtain patient-level outcomes (i.e. no patient cohort); an educational didactic session was added at the beginning of the intervention period; and providers received ≥2 academic detailing sessions throughout the intervention (vs. a single session in TOPCARE). In addition, reports that served as the basis for academic detailing from the electronic registry were updated to include clinic-wide averages for comparison, and summary tables of patients for each provider. Descriptions follow of the individual components of TEACH in order of implementation throughout the 12-month intervention window.

Didactic session

Within a few weeks of randomization, intervention providers received a 60-minute group didactic session from an expert on opioid prescribing.⁶⁹ Participants who were unable to join in person were given the opportunity to attend a live session via video conference. A case study tailored for each site was presented to engage providers in finding signs of nonmedical prescription opioid use, and devising treatment plans for complex situations with challenging patients. The opioid prescribing expert explained how to detect the non-medical use of prescription opioids, developed providers’ patient communication skills around this topic, and reviewed the study resource website (www.mytopcare.org) and brochure outlining opioid prescribing guidelines.

Collaboration with nurse care manager (NCM)

Each site hired a NCM with a background in HIV care. Both NCMs collaborated with providers in the intervention group throughout the year-long study period to conduct essential elements of guideline-driven care. They were introduced to the providers during the didactic sessions, and subsequently met with each provider to review their panel of patients receiving LTOT. The NCMs implemented opioid treatment agreements, UDTs, pill counts, and periodic checking of online prescription monitoring programs (PMPs), either directly or by assisting the provider depending on state regulations. The NCMs conducted an initial intake assessment for each patient, which included the following: a detailed medical, substance use, and social history; the current opioid misuse measure (COMM);³⁹ opioid risk tool (ORT);⁷⁹ and discussing and signing a treatment agreement. The NCM then followed-up regularly with patients regarding refills, pain assessments, UDTs, and pill counts, the frequency of which was determined by the patient’s risk level. Prior to the starting the intervention, the NCMs received extensive training from TOPCARE study nurses and study investigators in best practices to monitor opioids, deliver addiction care, and utilize the registry. During the intervention, the NCMs had the opportunity to discuss case management during weekly calls with co-investigators including opioid prescribing experts, an addiction specialist, and HIV providers.

Registry

The NCM utilized a web-based registry developed by the TOPCARE clinical research study team to record and view individual or aggregate information on opioid treatment agreements, UDTs, pill counts, and checking of online PMPs.⁶⁸ The registry was used for clinical purposes, and the data were not used in the RCT analysis. The registry also enabled the NCM to anticipate refills, ensure appropriate monitoring measures, and schedule follow-up appointments. The NCM regularly entered data into the registry at or after meetings with LTOT patients, and the registry would generate a note that could be easily copied and pasted into a patient encounter note in the electronic medical record. The registry also generated summary reports that were used in individual academic detailing sessions with intervention providers, to highlight high-risk LTOT patients. The following updates were made to the TOPCARE registry before initiating TEACH: adding an overview of all patients and their pertinent characteristics related to safe opioid prescribing (COMM score, UDTs), as well as a comparison of the provider’s panel to an aggregate of the providers in the rest of clinic.

Academic detailing

Intervention providers participated in two 30-minute individual academic detailing sessions throughout the year-long intervention, and were given the option of a third, booster session if desired. Session 1 was conducted by the opioid prescribing expert, the site’s NCM, and a co-investigator at each site who was designated the TEACH clinic leader, neither of whom specialized in opioid prescribing or addiction. Sessions 2 and 3 were conducted by the NCM and the TEACH clinic leader. At the academic detailing sessions, providers received a personalized folder prepared by the study staff and NCMs with registry-generated data on their LTOT patient panel (e.g. opioid risk assessment scores, completion of UDTs, and checking of online PMP). Electronic registry data provided clinic-level data on all the patients in the intervention as well as a summary of the provider’s panel. The NCM assisted in these sessions by providing specific cases and experiences from working with the patients. Providers were encouraged to utilize academic detailing sessions to discuss difficult cases and develop treatment plans with input from the intervention team.

Facilitated access to addiction specialist

The NCM encouraged and arranged referral of challenging patients with substance use disorders to addiction specialists affiliated with the HIV clinics when available. The study protocol did not dictate how such patients were managed; clinical decision-making remained in the hands of the provider, and treatment resources differed by site. In some cases, patients were transitioned from LTOT to buprenorphine for opioid use disorder, and the NCM continued oversight of treatment. As mentioned previously, a weekly meeting took place with study team investigators focused on the intervention, and provided an opportunity for the NCM to obtain regular assistance on any difficulties that had arisen, and the meetings took place throughout the duration of the intervention at both sites.

Quality assurance

The following measures were implemented in order to ensure the fidelity of the intervention. Prior to enrollment at each site, training occurred with the TEACH clinic leaders and NCMs via site visits, conference calls, and webinars. In addition, the TEACH NCMs shadowed the NCMs from the TOPCARE study. TEACH NCMs served as a resource for one another as well. Checklists were developed for academic detailing sessions to ensure that they were carried out consistently. During the weekly intervention meetings, the NCM filled out checklists documenting their interactions with intervention providers and TEACH clinic leaders documented any challenges the NCM had encountered.

Patient cohort

Cohort participants and recruitment

PLWH receiving LTOT (n = 170) were recruited into the TEACH cohort from July 2015 to December 2016 from two safety-net, hospital-based HIV clinics in Boston and Atlanta. Of the 170 participants recruited into the cohort, 117 were simultaneously enrolled in the RCT (Figure 1). Potential patient cohort participants were identified by queries of the electronic medical record by the clinical data warehouse (CDW) at BMC, and the Center For AIDS Research (CFAR) at Emory University. Initial patient participant inclusion criteria included: (1) ≥18 years of age, (2) diagnosis of HIV infection by ICD-9 codes or lab tests, (3) having received ≥3 opioid prescriptions ≥21 days apart within a 6-month period in the prior year, and (4) having attended ≥1 visit to the medical center’s enrollment sites within the prior 18 months. Study staff and clinician investigators reviewed the queried lists extensively to confirm that those included were eligible for the study. The lists from each site were generated a second time near the end of the recruitment period to ensure the inclusion of all potentially eligible patients.

Research staff checked the HIV clinic schedule daily using the EMR and approached potentially eligible patients for screening after informing their clinical team. Those who did not frequently come in for appointments were contacted by phone or letters at other hospital appointments to invite them to participate. Patients were invited to participate in the observational cohort regardless of whether their provider enrolled in the RCT. If the patient was interested, the RA scheduled an appointment to meet. At the scheduled meeting, the RA formally screened the patient and assessed secondary eligibility criteria. As the initial identification of potential participants was conducted using EMR data, we conducted an additional screening for the following: (1) provision of contact information of two individuals to assist with follow-up, (2) possession of a home or mobile telephone, and (3) English speaking. Secondary exclusion criteria included: (1) plans to move from the area within 12 months, and (2) inability to consent or understand interviews. The screener collected information on age, gender, race, ethnicity, and language. If eligible and willing, research staff obtained written informed consent, recorded contact information, administered the baseline assessment, and provided participant compensation. TEACH patient cohort enrollment is outlined in Figure 3.

Figure 3 Study flow diagram describing recruitment of patients living with HIV into the targeting effective analgesia in clinics for HIV (TEACH) patient cohort.

Patient cohort research assessments

Patient cohort participants underwent 60- to 90-minute assessments administered by a research assistant (RA) at baseline and 12-month follow-up (Table 2). The 12-month assessment was similar to the baseline assessment, with minor modifications to demographics, pain treatments, satisfaction with pain treatment, health literacy, and aberrant medication use (ORBIT assessment added at 12 months but not included at baseline⁸⁰). Participants were compensated with $35 upon completion of the baseline assessment and $50 upon completion of the 12-month follow-up assessment. Patient assessment data was entered directly into REDCap and a study tracking system by the RA when interviewing patients. RAs reviewed all data at the end of each assessment, and all data were double-checked by an additional staff member at each site for quality assurance. RAs contacted study participants at 3 months, 6 months, and 9 months after baseline to verify contact information and remind them of the 12-month follow-up interview. Participants received birthday and holiday letters with giftcards to a coffee shop ($2) throughout the year thanking them for their participation.

Table 2 TEACH patient cohort assessment components

1. Demographics^82

2. HIV transmission and diagnosis, ART, and HCV Testing^82

3. Brief pain inventory (BPI)^83–85

4. Pain treatments^33^,^34

5. LTOT providers and pharmacies^33^,^34

6. Patient satisfaction with pain management^33^,^34^,^86^,^87

7. Physician trust scale*^88–90

8. Aberrant opioid ue (COMM, ORBIT**)^39^,^91–93

9. Drug and alcohol use (AUDIT, ASI, TCU)^82^,^94–97

10. Service utilization^82^,^98

11. State trait anxiety inventory (STAI)^99^,^100

12. Depression (CES-D)^101^,^102

13. PTSD (PCL-5)^103

14. Overdose and suicide

15. Health literacy^§Citation82,^104

16. General health (VR-12)^105

17. Social support^106–108

*Wording changed from “physician” to “provider”.

^** At follow-up only.

^§ At baseline only.

EMR data extraction

At the conclusion of the study intervention period, the CDW and CFAR teams extracted data from the EMR for both the RCT and the cohort at BMC and Emory, respectively. Extracted information included UDTs, medications, HIV viral load, CD4 count, and comorbidities.

To ensure quality data, the data management team and CFAR staff conducted thorough checks comparing data to the EMR from which they were pulled. For each outcome variable, 10% were checked for accuracy. Discrepancies were logged and discussed, and algorithms for data extraction were refined to fix the source of the error. To ensure we did not include duplicate medications, we manually checked all medications that could potentially be duplicates to determine if they should be counted or deleted. We considered medications to be potential duplicates if two prescriptions were for the same medication, had the same description, and if the ordering dates were less than 7 days from one another. The data management team manually reviewed and discussed the list of potential duplicates to identify actual duplicates or abnormal situations that the algorithm may not have identified.

Manual EMR review

In addition to the EMR extraction, research staff conducted manual EMR reviews at the conclusion of the study, as some data elements could not be accurately captured by extraction. This manual review included information on opioid treatment agreements, discontinuation of opioids and reasons, emergency department utilization, pill counts, evidence of and/or treatment for opioid use disorder, and other substance use/mental health disorder diagnoses. A standard operating procedure manual was created to ensure uniform recording of these variables, and staff conducting the medical chart reviews met with the rest of the research team and study investigators to resolve any questions.

Data management

RCT and cohort data were managed by the Biostatistics and Epidemiology Data Analytics Center at Boston University School of Public Health and research staff at BMC. The cohort’s project website and study data were located on a secure server within the Boston University Medical Center (BUMC) domain. Access to the system was protected via secure logins and all data transmissions were encrypted using secure socket layering (SSL). All web-forms were protected using SSL encryption technology and files were protected by electronic firewalls that restricted access to designated users. Identifiers needed to track participants were kept separate from research data.

Measures

The primary outcome for Aim 1, improving HIV providers’ adherence to LTOT guidelines, will be measured by occurrence of ≥2 UDTs between the providers’ randomization date and 12-months. In the event that a confirmatory test is conducted to verify the results of a toxicology test, it will be considered part of the original test rather than a second test.

The primary outcome for Aim 2, improving patient-level outcomes, will be the percent of patients with any early refills between the providers’ randomization date and 12 months. Each opioid prescription will be analyzed using the instructions and total number of doses in order to determine an expected prescription duration in days, assuming maximum number of pills taken per day. Early refills are defined as prescriptions that are filled more than 3 days prior to the expected refill date.

The primary outcome for Aim 3, improving HIV providers' satisfaction with prescribing LTOT, will be the mean satisfaction score in the intervention group compared to the control group at the 12-month assessment. The question used is, “How satisfied are you in managing chronic opioid therapy in your HIV-infected patients who are on chronic opioid therapy for pain?” with responses ranging from 1 – “not at all” to 10 – “extremely”.

The primary outcome for Aim 4, improving virologic control among patients, will be the percent of patients with undetectable HIV viral load (<200 copies/mL) at the test closest to 12-month follow-up.

Analytic methods

Descriptive statistics will be calculated for patient-specific and provider-specific variables at baseline and the 12-month follow-up. At baseline, all variables will be assessed to ascertain important differences across the two randomized arms. Spearman correlation coefficients will be obtained to identify pairs of possible collinear variables (r > 0.4) and would therefore not be included together in regression analyses. In addition, the variance inflation factor (VIF) will be assessed for each covariate and randomization group to detect possible correlation. This study will use an intent-to-treat analysis including all participants according to their randomized assignment.

Randomization and the intervention occur at the provider level while the unit of observation includes both the individual patients receiving LTOT and the providers. Thus, analyses of patients must account for clustering by the provider. The primary analysis, evaluating the effect of the intervention on the binary study outcomes, will use generalized estimating equations (GEE) and logistic regression models with empirical standard errors to account for clustering by providers. The models will include the randomization group as the main independent variable and control for the randomization’s stratification factors, site, and provider volume in order to improve efficiency. In addition, the models will control for any important baseline provider or patient characteristics that differ between groups in order to avoid confounding. Potential patient-specific confounders of interest include demographics, age, gender, and hepatitis C status. Potential provider-specific covariates include those provider characteristics such as age, gender, and provider type (e.g. physician vs. advanced practice provider). For the outcomes of provider satisfaction and confidence at 12 months, we will use multiple linear regression models. If the distribution is skewed, transformations will be considered or a median regression model will be used. The secondary outcome, number of early refills, will be analyzed as count data using Poisson regression; if the variance exceeds the mean, a negative binomial regression model will be used. Analyses will be conducted using SAS version 9.4 (SAS Institute, Inc.).

We will test for differences in baseline characteristics between participants lost to follow-up and those who are not. Missing data patterns will be evaluated, including the percentage of those missing for each variable and the distribution of the number of variables missing for subjects. We will use multiple imputation (with 25 generated complete datasets) to account for missing outcome data. Variables used for imputation of patient-level outcomes will include gender, age, depression, hazardous drinking, drug use, race, Hispanic, BMI, CD4 cell count, HIV viral load, Charlson Comorbidity Index,⁸¹ randomization group, stratification variables, and baseline value of the outcome. For patient-level outcomes, data will be imputed separately by provider. Variables used for imputation of provider-level outcomes included gender, years of practice in HIV care, patient volume, site, randomization group, and baseline value of the outcome. All analyses will be conducted using SAS version 9.3.

Power calculations

Power calculations to define the limits of the study were conducted as follows. The calculations assumed a two-sided test, with a significance level of 0.05. It was expected that 35 total physicians would be enrolled in the study, with an average of five eligible patients per physician for a total of 175 patients. We expected the intraclass correlation coefficient (ICC) will be <0.10 for the outcomes of interest, and conservatively assume a value of 0.10 in the following calculations. Calculations for binary outcomes were based on a chi-square test with continuity correction and continuous outcomes are based on the two sample t-test, with estimates adjusted for clustering based on the inflation factor (also referred to as the design effect). For the outcome ≥2 UDTs, the minimum detectable difference is 28% (i.e. 26% vs. 54% in the control and intervention groups, respectively). For the outcome any early refills, the minimum detectable difference is 27% (i.e. 46% vs. 19% in the control and intervention groups, respectively). For the outcome physician satisfaction, the minimum detectable difference is 2.5 in mean satisfaction scores (e.g. 7.1 vs. 4.6 for the intervention and control groups).

Discussion and impact

The TEACH study tests the effectiveness of a collaborative care intervention on HIV providers’ opioid prescribing for chronic pain. The intervention seeks to increase providers’ adherence to current guidelines for care with the use of LTOT, prevent nonmedical use of prescription opioids among patients, and heighten physician and patient satisfaction in regard to this dimension of care. The TEACH study will provide patient and provider-level effects of an intervention to improve opioid prescribing for chronic pain in HIV-clinics. The results from this cluster randomized trial design should inform delivery of care for PLWH who are on LTOT for chronic pain and provide a “blueprint” for dissemination.

Clinical trial registration details

These studies were registered with ClinicalTrials.gov through the National Institutes of Health – Targeting Effective Analgesia in Clinics for HIV – Intervention, NCT02564341; and Targeting Effective Analgesia in Clinics for HIV – Patient Cohort, NCT02525731.

Ethics approval and consent to participate

The TEACH study was approved by the institutional review boards at Boston University Medical Campus and Emory University.

Notes on Contributors

Marlene C. Lira, is a Research and Education Project Manager at Boston Medical Center, and manages clinical and policy research related to substance use, as well as training initiatives about addiction for physicians. She is currently pursuing her MPH in Epidemiology at the Harvard T.H. Chan School of Public Health.

Judith I. Tsui, M.D., M.P.H., is a board certified physician at Harborview and a University of Washington Associate Professor of General Internal Medicine. Her research has elucidated complications of substance abuse and related viral infections, as well as the positive impact of addiction treatment on hepatitis C.

Jane M. Liebschutz, MD, MPH is Chief of General Internal Medicine at University of Pittsburgh Medical Center. Her previous research, including the TOPCARE study, has focused on safe opioid prescribing.

Jonathan Colasanti, MD, MPH is an Assistant Professor of Medicine in the Division of Infectious Diseases at Emory University.

Christin Root, is the Associate Director of Programs in the Hubert Department of Global Health and the Rollins School of Public Health at Emory University.

Debbie M. Cheng, ScD, is Professor of Biostatistics and has been on the faculty at the Boston University School of Public Health since 2002. She collaborates with investigators at the Boston University School of Medicine on numerous projects in the areas of substance abuse and HIV research.

Alexander Y. Walley, M.D., M.Sc., is an Associate Professor of Medicine at Boston University School of Medicine and the Director of the Boston Medical Center Addiction Medicine Fellowship program. He provides primary care and office-based addiction treatment for patients with HIV at Boston Medical Center and methadone maintenance treatment at Health Care Resource Centers.

Meg Sullivan, MD was the Clinical Director of HIV Services at Boston Medical Center (BMC). She has focused on models for successful engagement and retention of HIV positive patients into ongoing medical care.

Christopher Shanahan, MD, MPH, is an Assistant Professor at Boston University School of Medicine. He is the Director of the Community Medicine Unit within the Section of General Internal Medicine at Boston University School of Medicine. He was the former Associate Medical Director for the Massachusetts Screening Brief Intervention and Referral to Treatment Program.

Kristen O'Connor, BCN, was a TEACH Nurse Care Manager at Boston Medical Center.

Catherine Abrams, BSN, was a TEACH Nurse Care Manager at Grady.

Leah S. Forman, MPH, is a Statistical Programmer and Data Manager in the Biostatistics and Epidemiology Data Analytics Center at Boston University School of Public Health.

Christine Chaisson, MPH, is the Director of Translational Research at Optum Labs. She previously served as the Director of the Biostatistics and Epidemiology Data Analytics Center at the Boston University School of Public Health.

Carly Bridden, MA, MPH is the Clinical Research Director for the Clinical Addiction Research and Education (CARE) Unit at Boston Medical Center.

Melissa C. Podolsky, was the Project Coordinator on the TEACH study at Boston Medical Center. She is currently pursuing her MPH at the Rollins School of Public Health at Emory University.

Kishna Outlaw is a Research Interviewer at Emory University.

Catherine E. Harris is a Research Interviewer in the Department of Global Health and post-baccalaureate student at the Grady Nia Project in the Department of Psychiatry and Behavioral Sciences at Emory.

Wendy S. Armstrong, MD, FIDSA, FACP is the Vice Chair of Education and Integration and Professor of Medicine in the Division of Infectious Diseases at Emory University School of Medicine. She has been active in the Emory Center for AIDS Research and established the Emory CFAR HIV Specimen Repository.

Carlos del Rio, MD is the Hubert Professor and Chair of the Department of Global Health and Professor of Epidemiology at the Rollins School of Public Health and Professor of Medicine in the Division of Infectious Diseases at Emory University School of Medicine. He is also co-Director of the Emory Center for AIDS Research (CFAR).

Jeffrey H. Samet, MD, MA, MPH, is Chief of General Internal Medicine and the John T. Noble Professor of Medicine at Boston University Schools of Medicine and Public Health. His research focuses on the intersection of HIV and substance use.

Acknowledgements

The authors thank the HIV providers and their patients who enrolled in this study. We are grateful to Linda Rosen at Boston Medical Center’s Clinical Data Warehouse, and Minh Nguyen and Jeselyn Rhodes of the Emory Center For AIDS Research, for their assistance with data extraction.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The project described was supported by the National Institute on Drug Abuse under grant 5R01DA037768-05; the Center for AIDS Research at Emory University under Grant 5P30AI050409-20; and the Providence/Boston Center for AIDS Research under grant 2P30AI042853-20A1. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Drug Abuse or the National Institutes of Health.

Notes on contributors

Marlene C. Lira

Marlene Lira, BA, is Research and Education Project Manager at Boston Medical Center, and manages clinical and policy research related to substance use, as well as training initiatives about addiction for physicians. She is currently pursuing her MPH in Epidemiology at the Harvard T.H. Chan School of Public Health.

Judith I. Tsui

Judith Tsui, M.D., M.P.H., is a board certified physician at Harborview and a University of Washington Associate Professor of General Internal Medicine. Her research has elucidated complications of substance abuse and related viral infections, as well as the positive impact of addiction treatment on hepatitis C.

Jane M. Liebschutz

Jane Liebschutz, MD, MPH is Chief of General Internal Medicine at University of Pittsburgh Medical Center. Her previous research, including the TOPCARE study, has focused on safe opioid prescribing.

Jonathan Colasanti

Jonathan Colasanti, MD, MPH is Assistant Professor of Medicine in the Division of Infectious Diseases at Emory University and Associate Medical Director of the Infectious Diseases Program at the Grady Health System.

Christin Root

Christin Root, is the Associate Director of Programs in the Hubert Department of Global Health and the Rollins School of Public Health at Emory University.

Debbie M. Cheng

Debbie M. Cheng, ScD, is Professor of Biostatistics and has been on the faculty at the School of Public Health since 2002. She collaborates with investigators at the Boston University School of Medicine on numerous projects in the areas of substance abuse and HIV research.

Alexander Y. Walley

Alexander Y. Walley, M.D., M.Sc., is an Associate Professor of Medicine at Boston University School of Medicine and director of the Boston Medical Center Addiction Medicine Fellowship program.

Meg Sullivan

Meg Sullivan, MD is the Clinical Director of HIV Services at Boston Medical Center (BMC). She has focused on models for successful engagement and retention of HIV positive patients into ongoing medical care.

Christopher Shanahan

Christopher Shanahan, MD is an Assistant Professor at Boston University School of Medicine. He is the Director of the Community Medicine Unit within the Section of General Internal Medicine at Boston University School of Medicine. He was the former Associate Medical Director for the Massachusetts Screening Brief Intervention and Referral to Treatment Program.

Kristen O’Connor

Kristen O’Connor, BCN, is a TEACH Nurse Care Manager at Boston Medical Center.

Catherine Abrams

Catherine Abrams, BSN, is a TEACH Nurse Care Manager at Grady.

Leah S. Forman

Leah Forman, MPH, is a Statistical Programmer and Data Manager in the Biostatistics and Epidemiology Data Analytics Center at Boston University School of Public Health.

Christine Chaisson

Christine Chaisson, MPH, is the Director of Translational Research at Optum Labs. She previously served as the Director of the Data Coordinating Center at the Boston University School of Public Health.

Carly Bridden

Carly Bridden, MA, MPH is the Clinical Research Director for the Clinical Addiction Research and Education (CARE) Unit at Boston Medical Center.

Melissa C. Podolsky

Melissa Podolsky, BA is the Project Coordinator on the TEACH study at Boston Medical Center. She is currently pursuing her MPH at the Rollins School of Public Health at Emory University.

Kishna Outlaw

Kishna Outlaw is a Research Interviewer at Emory University.

Catherine E. Harris

Catherine Harris is a Research Interviewer in the Department of Global Health and post-baccalaureate student at the Grady Nia Project in the Department of Psychiatry and Behavioral Sciences at Emory.

Wendy S. Armstrong

Wendy Armstrong, MD, FIDSA, FACP is the Vice Chair of Education and Integration and Professor of Medicine in the Division of Infectious Diseases at Emory University School of Medicine. She is the Medical Director of the Grady Infectious Diseases Program and is an active investigator of the Emory Center for AIDS Research having established the Emory CFAR HIV Specimen Repository.

Carlos del Rio

Carlos del Rio, MD is the Hubert Professor and Chair of the Department of Global Health and Professor of Epidemiology at the Rollins School of Public Health and Professor of Medicine in the Division of Infectious Diseases at Emory University School of Medicine. He is also co-Director of the Emory Center for AIDS Research (CFAR). His research focuses on improving outcomes for Persons Living with HIV with a particular focus on persons who use drugs.

Jeffrey H. Samet

Jeffrey Samet, MD, MA, MPH, is Chief of General Internal Medicine and the John T. Noble Professor of Medicine at Boston University Schools of Medicine and Public Health. His research focuses on the intersection of HIV and substance use.

Appendix