Clinics Optimizing MEthadone Take-homes for opioid use disorder (COMET): Protocol for a stepped-wedge randomized trial to facilitate clinic level changes

Abstract

Introduction

Regulatory changes made during the COVID-19 public health emergency (PHE) that relaxed criteria for take-home dosing (THD) of methadone offer an opportunity to improve quality of care with a lifesaving treatment. There is a pressing need for research to study the long-term effects of the new PHE THD rules and to test data-driven interventions to promote more effective adoption by opioid treatment programs (OTPs). We propose a two-phase project to develop and test a multidimensional intervention for OTPs that leverages information from large State administrative data.

Methods and analysis

We propose a two-phased project to develop then test a multidimensional OTP intervention to address clinical decision making, regulatory confusion, legal liability concerns, capacity for clinical practice change, and financial barriers to THD. The intervention will include OTP THD specific dashboards drawn from multiple State databases. The approach will be informed by the Health Equity Implementation Framework (HEIF). In phase 1, we will employ an explanatory sequential mixed methods design to combine analysis of large state administrative databases—Medicaid, treatment registry, THD reporting—with qualitative interviews to develop and refine the intervention. In phase 2, we will conduct a stepped-wedge trial over three years with 36 OTPs randomized to 6 cohorts of a six-month clinic-level intervention. The trial will test intervention effects on OTP-level implementation outcomes and patient outcomes (1) THD use; 2) retention in care; and 3) adverse healthcare events). We will specifically examine intervention effects for Black and Latinx clients. A concurrent triangulation mixed methods design will be used: quantitative and qualitative data collection will occur concurrently and results will be integrated after analysis of each. We will employ generalized linear mixed models (GLMMs) in the analysis of stepped-wedge trials. The primary outcome will be weekly or greater THD. The semi-structured interviews will be transcribed and analyzed with Dedoose to identify key facilitators, barriers, and experiences according to HEIF constructs using directed content analysis.

Discussion

This multi-phase, embedded mixed methods project addresses a critical need to support long-term practice changes in methadone treatment for opioid use disorder following systemic changes emerging from the PHE—particularly for Black and Latinx individuals with opioid use disorder. By combining findings from analyses of large administrative data with lessons gleaned from qualitative interviews of OTPs that were flexible with THD and those that were not, we will build and test the intervention to coach clinics to increase flexibility with THD. The findings will inform policy at the local and national level.

Introduction

Methadone is a highly effective treatment for opioid use disorder (OUD) that has been available since the 1960s [1–7]. In the United States, methadone is provided through clinics called opioid treatment programs (OTPs) that are highly regulated by the Drug Enforcement Agency (DEA), the Substance Abuse and Mental Health Services Administration (SAMHSA), and state and local governments. There are over 1,800 OTPs in the United States [8]; however, the regulations and the OTP system that have emerged since the Vietnam War era reflect stigmatizing and discriminatory notions of people with OUD (PWOUD) that influence clinical practices to this date [9–15].

Prior to the COVID-19 PHE, SAMHSA regulations required the majority of OTP clients to visit clinics almost daily to monitor their methadone dosing [16, 17]. Clients could eventually earn privileges for THD that would allow for self-administration outside the clinic setting, but the bar was high. A client needed to be in treatment for 9 months before being eligible for weekly THD and 12 months before being eligible for bimonthly THD. In addition to the time in treatment criteria, clients need to be abstinent from other substances, including alcohol and cannabis, be ‘compliant’ with treatment, and be relatively high functioning in other life domains [16, 18–22]. The SAMHSA 8-point criteria for compliance and ‘stability’ were not defined in detail and left open to OTP staff clinical discretion. Staff generally defaulted to risk-averse interpretations that were often not aligned with existing data [23]. As a result, clients were subjected to burdensome and demeaning treatment that created barriers to employment or improvements in other domains of life functioning [24–27]. The barriers are particularly troublesome for PWOUD who live in rural areas or with long travel distances to OTPs [28–31]. Notably, Black/African American clients were subject to more restrictive interpretations of the criteria than non-Latinx White clients [32]. The presumed rationale for the restrictive regulations was to mitigate risk of diversion or overdose from methadone [22]; however, the effect has been to limit access and place a heavy burden on clients that has been referred to as “liquid handcuffs [27]”.

In response to the COVID-19 Public Health Emergency (PHE), SAMHSA gave states a temporary option to request a waiver to allow greater flexibility of take-home dosing (THD) of methadone for clients [33]. Thus, the PHE federal rules [33] shifted from time-in-treatment requirements to clinical judgement regarding patient stability. SAMHSA issued an extension of the COVID-19 THD waiver option for states indefinitely [33]. Uptake by states of the new waiver option has been uneven. Not all states applied for the waiver [34, 35], which reflected the variability in local regulatory approaches [36]. SAMHSA recently published a notice of proposed rulemaking to make the flexibility of THD permanent [37]. The response by OTP organizational leadership and staff has been mixed. While some have highlighted the benefits to clients of greater flexibility in THD, others have expressed concern about the potential harm to clients from unsupervised dosing that may result in overdoses, diversion, or program legal liability from potential misuse [21, 38–42].

There is no evidence that greater THD is associated with more overdoses from methadone [43–48]. A recent large and robust study from Ontario, Canada, found that more flexible THD increased retention in care yet was not associated with increased mortality [49]. Cross-national comparisons with other countries that have less restrictive regulations do not demonstrate higher mortality or diversion than in the United States [50–56]. DEA reports show little evidence of large-scale diversion of methadone dispensed through OTPs, and recent studies show that mortality and diversion have been more strongly associated with methadone prescribed for pain management than for OUD [57–59]. Publications based on recent experience from the COVID-19 PHE by two large OTP providers in New York City have reported positive responses from clients, general support from OTP staff, and no evidence of significant increases in overdose or diversion [60, 61].

In New York, as elsewhere in the United States, the adoption of THD by OTPs across the state has been uneven. There is a pressing need for research to study the long-term implications of the PHE THD rules as well as to test data-driven interventions to promote more effective decision making and clinical practice regarding THD by OTPs [22, 62]. We are conducting a two-phase project to develop and test a multidimensional intervention for OTPs that leverages information from large State administrative databases. By combining findings from analyses of large administrative data with lessons gleaned from qualitative interviews of clinics that have implemented THD, we will build scientifically supported knowledge to inform policy at a larger system level. Our protocol is informed by the Health Equity Implementation Framework (HEIF) to guide the development and study of an intervention to increase THD among OTPs. HEIF integrates two research frameworks for conducting implementation studies of interventions targeting underserved populations: 1) the Kilbourne framework for health disparities research; and 2) the Integrated Promoting Action on Research Implementation in Health Services (i-PARIHS) framework. The Kilbourne framework focuses on historical, cultural, and contextual factors that affect client-provider encounters [63]. i-PARIHS posits that optimal implementation occurs when practice facilitation promotes the acceptance and use of a new practice innovation by tailoring it to the recipient’s specific needs [64–66]. Facilitators are the active ingredient that help navigate individuals and teams through complex change processes by addressing: a) the innovation’s degree of fit within the existing practice; b) the motivations, beliefs, goals, characteristics, and resources of the intervention recipients; and c) the inner and outer context in terms of leadership support, culture, past innovation experiences, the learning environment, organizational priorities, capacity for change, regulatory/policy drivers, incentives/mandates, and system stability/instability.

The COMET: Clinics Optimizing MEthadone Take-homes for opioid use disorder study is a two-phase project to develop then test a multidimensional OTP intervention to address clinical decision making, regulatory confusion, legal liability concerns, capacity for clinical practice change, and financial barriers to THD.

We have six main aims:

Phase 1, Year 1: Intervention development

1. Analyze administrative data to identify factors associated with OTP variation in THD practices and categorize clinics by THD flexibility.

2. Conduct qualitative interviews with leadership and staff of 10 OTPs (5 high and 5 low THD flexibility) on clinical and organizational factors affecting take-home dosing decisions.

3. Complete development of the multidimensional OTP intervention.

Phase 2, Years 2–5: Stepped-wedge trial

1. Test the effects of the intervention on THD, retention in care, and adverse events.

2. Assess clinic implementation outcomes and conduct qualitative interviews with OTP leadership, staff, and clients on attitudes, experiences and behaviors related to the intervention.

3. Using mixed methods, explore variation in THD associated with race and ethnicity.

Multiple systemic reviews have concluded that addiction treatment in the United States has large gaps in quality of care and limited capacity for clinical program improvement, and there are no agreed upon quality metrics for the SUD treatment field [67–72]. Notably, the workforce has limited levels of education or professional training to implement process change initiatives to improve clinical outcomes [67, 70, 71, 73]. Because of this limitation, the proposed intervention will draw from our current research [74] based on principles of process improvement and the iPARIHS framework to coach OTPs on changing protocols and workflows to adopt more flexible THD [75–82]. One obstacle to adopting new THD practices is uncertainty by the senior leadership of OTPs regarding the financial viability of any change that deviates from established billing practices [60]. As part of this project, we will use State administrative data to develop a revenue projection tool for OTPs; this decision-support tool will be provided to OTP leadership and allows for revenue projections conditional on target THD practices and optimal billing [83, 84].

Participant and data confidentiality

Every effort will be made to ensure participant confidentiality. Leadership at participating clinics will not be informed about any information regarding participation in the study. All of the personnel and staff on the study team have been certified in NYU’s Human Research Social/Behavioral Research Course through the CITI program. The study team recognizes that the protection of human subjects relies on policies that secure data with personally identifiable information. Each of the investigators has adopted, and will strictly follow, such policies for all data used in the study. Any time data is reported, it will be aggregated with that of other respondents (e.g., to discuss identified themes or key concepts) and never reported with personally identifiable information. No information will be reported that could make it possible for anyone to identify participants in any presentations or written reports about this study. Any identifiable personal information that participants may reveal about themselves during the interviews will be removed from the transcripts. If a direct quote from an interview is utilized in a report or publication (e.g., to exemplify a concept that arose in multiple interviews), it will only be cited using a pseudonym or participant ID number. All information will be stored in our secure, encrypted NYU drives. Personal identifying information will never be shared with anyone outside of the research team. In the event that a subject revokes authorization to collect or use PHI, the investigator, by regulation, retains the ability to use all information collected prior to the revocation of subject authorization. For subjects that have revoked authorization to collect or use PHI, attempts should be made to obtain permission to collect at least vital status (i.e. that the subject is alive) at the end of their scheduled study period.

Ethics approval

Ethics approvals have been obtained using a single IRB from NYU Langone Health (#S22-00892), overseeing all sites participating in the study: New York State Office of Addiction Services and Supports (OASAS), University of Connecticut, and Weill Cornell Medicine. The study was registered under Clinicaltrials.gov (NCT05675735), and the SPIRIT checklist was used as a guide for reporting this study protocol (Fig 1).

Fig 1. SPIRIT flow diagram for the COMET study.

Material and methods

Phase 1. Intervention development

Aim 1. Analyze administrative data to identify factors associated with OTP variation in THD practices and categorize clinics by THD flexibility

We will join multiple administrative data sources for years March 2020 to February 2021 to form our analytical dataset. An OASAS treatment registry (Client Data System, CDS) will provide extensive clinical data on clients at treatment admission. All licensed SUD treatment programs in NYS must enter admission and discharge data into the CDS. Medicaid will provide medical diagnoses, healthcare services, OTP visits, and methadone administration billing data. The OTP Opioid Treatment Annual Update Report (OTAU) database will provide weekly, bimonthly, monthly THD measures at clinics. The outcome measure is OTP-level THD use. We will descriptively assess variations in THD flexibility (1. no THD; 2. weekly supply; 3. bimonthly; 4. month supply) and rank clinics.

Aim 2. Conduct qualitative interviews with leadership and staff of 10 OTPs (5 high and 5 low THD flexibility) on clinical and organizational factors affecting take-home dosing decisions

We will select 10 clinics for qualitative interviews with staff members to explore clinic-level factors associated with THD practices and technical assistance needed to implement more flexible THD. Clinic staff inclusion will include anyone who works at the 10 clinics that the OASAS client data system generates from the quantitative analysis in year 1. The semi-structured 1:1 interviews will be recorded and transcribed and then using directed content analysis following steps recommended for content analysis [85–87]. We will use rapid analysis using matrices to analyze the data in real-time [88–90]. Researchers will make field notes immediately after interviews. Weekly meetings will be held to review coding decisions, discuss discrepancies, and check progress. As emerging concepts are identified, we will adapt the existing coding structure. When the analysis is complete, the team will meet to review summaries of the qualitative results and refine hypotheses about the contextual factors and strategies that lead to better outcomes and those that might be barriers to success.

Aim 3. Complete development of the multidimensional OTP intervention

We plan to convene an online stakeholder group meeting comprised of leadership from providers, health insurers, clients, and family members to provide input on qualitative research findings and subsequent intervention design.

Intervention structure

Table 1 provides an overview of the multidimensional OTP intervention and its goals. The intervention draws from a similar approach our team is currently employing to coach non-OTP SUD treatment clinics on quality improvement [74]. The proximal goal will be to increase OTP uptake of flexible THD, which in turn should result in increased retention in care without increasing adverse healthcare events or mortality. The intervention addresses facilitators and barriers to THD practices as identified by existing studies as well as those identified during phase 1 of the current project. Guided by HEIF and research on organization change [91–95], the intervention will be designed to address the information gaps, training needs, and beliefs of individuals across the organizations. Senior leadership have purview over financial and legal matters. Medical directors and nursing staff are responsible for dosing decisions consistent with best clinical practices. Counseling and administrative staff are involved in multiple dimensions of clinic operations. Table 1 describes the components of the intervention as well as the audience within the OTP for each. It aligns with the HEIF by addressing clinical encounter factors (e.g., THD best practices), inner and outer context (e.g., regulatory guidance, data feedback), and economic influences (e.g., revenue projection tool). A designated external facilitator will work closely with each clinic using remote conferencing capabilities, as in our current trial. These facilitators will be trained in quality improvement practices and data management as well as relevant clinical topics (e.g., THD best practices). They will be supervised in weekly intervention oversight meetings by study team members. Facilitators will work closely with an identified member of clinic leadership, a clinical champion, and a small implementation team (e.g., 4–6 staff).

Table 1. COMET intervention.

Component	Who	Description
Data Dashboard	All	Performance feedback is an organizational intervention with robust scientific support [91–93]. Drawing from analytical methods applied during phase 1 of the project, provide clinic-level reports that show relative performance compared to other OTPs within the State on THD, retention, and adverse healthcare events. Relative performance will be adjusted for clients’ demographic and clinical characteristics [96–99]. The data feedback will specifically detail THD performance by race, ethnicity, and gender. Reports would be available for all staff.
Legal/Regulatory	All	To address barriers related to uncertainty over legal and regulatory risk [21, 38–42], provide guidance on Federal and State requirements for THD as well as guidance on managing legal liability risk. Guidance would be available for all staff.
Financial Guidance	SL	To address senior leadership concerns about financial viability, provide a revenue projection tool along with training to allow clinics to model out the financial effects of using new THD bundled payment billing. The tool and guidance will walk through number of clients with different THD schedules and projections for revenue based on optimal billing. Leadership would be able to conduct ‘what-if’ analyses under varying assumptions of THD practices [83, 84].
THD Best Practices	CL	Drawing from extant research as well as from findings from qualitative studies conducted during phase 1, provide training on clinical best practices for THD. The training would address parameters for determining level of client stability, protocols for monitoring client status, and use of technology [100–102]. The training will also highlight any clinic THD variation by race, ethnicity, or gender. The training would specifically address concerns and best practices for managing risks of methadone overdose and/or diversion. Our team will provide training to Medical Directors and nursing staff.
Process Change	C+A	Training and support on process improvement strategies drawn from organizational change management. The OTP will be asked to designate a clinic implementation team led by local champion to review data and guide practice change projects [67–72]. The project facilitator will train and support the implementation team on rapid cycle change strategies drawn from management science and used in our current clinic-level trial.

Table Notes: All = All OTP staff, including senior leadership, medical, counseling, and administrators; SL = senior leadership, including chief executive officer, financial, and clinic directors; CL = Clinical staff, including medical (physicians, nurses, nurse practitioners, physician assistants) and counselors; C+A = Clinical and administrative staff (e.g., reception, billing)

Phase 2. Stepped-wedge trial

We propose an embedded mixed methods design, which places the qualitative findings in context of the larger trial. We will assess outcomes quantitatively and answer the primary research questions, then use qualitative data to answer secondary questions. In years 2–5, we will employ a stepped-wedged randomized control trial with 36 OTPs chosen by the OASAS client data system which will be randomized into 6 cohorts of a six-month intervention.

Recruitment of sites

OTPs will be recruited to participate in this stepped-wedge trial by State partners via email invitations, announcements via State listservs, and announcements at relevant meetings. A total of 36 programs will be recruited at the beginning of the trial (Year 2). OTP that participated in the qualitative interviews during Phase 1 can also participate in the stepped-wedge trial. We will exclude clinics that already have very flexible THD policies (e.g., greater than 75% of clients are on weekly or greater THD schedule). The intervention is delivered to the leadership and staff members at the clinic-level. Zoom meetings will be conducted with directors of programs to confirm their enrollment and to explain the timeline and expectations. The study team will monitor enrollment and if enough clinics do not enroll, additional recruitment strategies will be discussed with the study team and implemented. In order to retain clinics in the trial, the study team will stay in regular contact with the clinic program directors. If a clinic is considering withdrawing from the study, the study team will schedule a meeting to further discuss and try to address reasons for desired withdrawal. We expect that offering the intervention to clinics will be attractive because the intervention will offer training and external facilitation free of charge. The study team has found in their experience that providing Statewide technical assistance is very welcome during the period of state reforms and the COVID-19 PHE; therefore, we do not anticipate problems with enrollment.

Aim 4. Test the effects of the intervention on THD, retention in care, and adverse events

Data. Similar to Phase 1 of the study, we will join multiple administrative data sources for years 2020 to 2027 to form our analytical dataset. We will use CDS, Medicaid claims, OTAU, and NYS vital statistics data to form the analytic dataset. The New York State vital statistics reporting will provide data on mortality for OTP clients. Our experience has been that accurate overdose data is reported with a long lag time due to capacity constraints for extensive forensic analyses across many of the Local Government Units (LGUs). Challenges in obtaining timely overdose data are not unique to New York and are a function of the complexity involved in assigning cause of death as well as local resources available to conduct these assessments [45, 103]. Consequently, we rely on obtaining all-cause mortality for clients in the OTPs participating in the trial. Other OASAS data will provide basic operational and staffing data for each of the OTPs. We have extensive experience joining these databases. Specifically, we have joined 83% of all OTP clients found in the OASAS registry to Medicaid claims data.

Measures. The outcome measures include: client-level THD use, retention in treatment, substance-related adverse events, all-cause adverse events, and all-cause mortality.

Client THD use will be estimated using billing data from Medicaid, identify clients receiving weekly, bimonthly, and monthly THD. Retention will be derived from the CDS and be the number of days until discharge. Substance-related adverse events are defined as the following: overdose, medically managed or supervised withdrawal services, hospitalizations or emergency department visits with a primary diagnosis of substance use disorder. All-cause adverse events will be estimated by examining any emergency department visits or hospitalizations irrespective of diagnosis. All-cause mortality will be defined as OTP client deaths irrespective of cause. As mentioned previously, we focus on all-cause mortality due to challenges with reliable data reporting on methadone and other opioid related overdose deaths [45, 103]. All-cause mortality will provide a broad indicator of health associated with longer retention in care as well as subsume any overdoses that may emerge from greater THD. A complete and detailed list of measures that will be used in the study can be found in Table 2.

Table 2. COMET study measures.

Domain	Variables	Source
Individual Baseline Characteristics
Demographic	Client age; race/ethnicity; gender; education; unhoused status; employment; criminal justice involvement	OASAS Medicaid
Clinical complexity	Mental health conditions (e.g., PTSD, ADHD); serious mental illness (SMI) (e.g., schizophrenia); chronic physical conditions (e.g., HIV, HCV)	Medicaid
Substance Use Patterns	Age of onset; heroin use; other substances used (alcohol, stimulants, cannabis, cocaine, and etc); frequency of use; and injection drug use (IDU)	OASAS
Prior Healthcare Utilization	History of use of outpatient, inpatient, emergency department	Medicaid
Individual-level Treatment Process
Psychosocial Interventions*	Behavioral treatment (i.e., group, individual, peer, or family counseling); telepractice; urine toxicology testing; mental health treatment; Medicaid reimbursed transportation	Medicaid
OTP Organizational Characteristics
Structure	Program size (i.e., annual patient volume); hospital system affiliation; number of OTPs within parent organization	OASAS Medicaid
Staffing	Number of staff; staff professionalization (e.g., degrees); gender and race/ethnicity; client-to-staff ratio	OASAS
Clinical Process	average monthly volume claims for individual, group, peer, and family counseling; average monthly volume claims for telepractice use; monthly volume for urine toxicology testing	Medicaid
Program THD*	OTP % of clients receiving weekly, bimonthly, and monthly THD	OASAS
Outcome Variables
Client THD*	Client THD in ordinal measure of weekly, bimonthly, and monthly dose	Medicaid
Retention*	Time until treatment exit	OASAS
Substance-related Adverse Event*	SUD-related emergency department visit or hospitalization, including for overdose; medically managed or supervised withdrawal services	Medicaid
Healthcare Utilization*	All-cause emergency department visits and hospitalizations	Medicaid
Mortality	Client level all-cause mortality	DOH

Table Notes: OASAS = CDS, OTP THD, and program databases; DOH = New York Department of Health Vital Statistics * Time-varying variables

Statistical power

To estimate power for select outcomes, we used a combination of PASS 2022 and Monte Carlo simulation [104]. Using administrative data to estimate sample sizes and baseline rates (μ), we compute detectable differences for 80% power, α = 0.05, ICC = 0.05, and SW-RCT with 6 crossover points, and 6 clinics randomized to start at each crossover point. For weekly or greater THD (i.e., picking up doses weekly or less often), we assume the current prevalence is 0.45. With 36 clinics and approximately 350 clients observed in each clinic in each of the seven periods (every six-month) (n≈88,200 total observations), power is 93% to detect a small increase (OR = 1.1) in the odds of weekly or greater THD. This corresponds to an increase from a prevalence of 0.45 under the control condition to 0.47 after intervention. Even when considering racial/ethnic patient subgroups making up just 10% of clinic clients (n≈35 per site and period), power is 92% to detect a modest increase in the odds of weekly or greater THD (OR = 1.3; an increase from 0.45 to 0.51). Because power is sufficient to detect small intervention effects for the binary outcomes of weekly or greater THD and adverse events, it is also sufficient to detect small intervention effects on retention in care.

Analysis of the stepped-wedge trial

Fig 2 depicts the stepped-wedge research design. Clinics are randomized to onset of intervention in one of six time periods over the three years of the intervention rollout (years 2–4 of the project) and tracked using administrative data. Data for all OTP clients during each of the seven six-month periods (n≈350/clinic) will be drawn from administrative data. The first six-month period will be a baseline prior to any intervention. Six clinics will be assigned randomly to one of six time periods for onset of intervention.

Fig 2. COMET implementation: Stepped-wedge design.

We will employ GLMM models commonly used in the analysis of stepped-wedge trials [105, 106]. The models will take the form:

$$F\left({\mu }_{ijk}\right)=\mu +{\beta }_j+{\delta X}_{ij}+{\alpha }_i+{\varphi }_{ik},{\alpha }_i\sim N\left(0,{\tau }_{\alpha }^2\right),{\gamma }_{ij}\sim N(0,{\tau }_{\varphi }^2)$$

where β_j is a fixed effect for time, X_ij is an indicator for intervention in clinic i at time j (coded 0 for the control condition, then 1 from when intervention begins in the clinic through end of study), and δ is the treatment effect. The models have random effects for clustering at the clinic and individual patient levels, which are assumed to be normally distributed with mean zero and variances ${\tau }_{\alpha }^2$ and ${\tau }_{\varphi }^2$, respectively. We will estimate the effect of intervention while controlling for secular trends and adjusting for clustering within clinic (α_i) and individual (ϕ_ik), using mixed effects modeling with binary distribution and logit link. Models will be fit using the glmmTMB package using R software [107, 108]. The primary outcome will be weekly or greater THD, and the fixed-effects coefficients for the intervention effect, exponentiated, will indicate how the intervention increases the odds of weekly or greater THD. Similar mixed models will consider other THD frequencies as outcomes (i.e., bimonthly and monthly). The same mixed logistic regression model will be used with any adverse events as the outcome, but a one-sided test of non-inferiority (i.e., adverse events are no worse under intervention than under the control condition) will be used [109]. For the retention outcome, a multilevel discrete-time survival model will be used [110–112]. An expanded person-period dataset will be constructed [113]. The outcome will have two possible states in any time period: 0 = retained in treatment; 1 = treatment exit. The multilevel analysis approach can accommodate recurring events for the same individual, with individuals leaving the set of patients “at risk” for dropout until treatment is initiated again. Analysis of the person-period dataset will use a mixed-effects regression model with a complementary log-log link function and random effects for patient and clinic. With the complementary log-log link, the coefficients can be interpreted as the relative effect on the hazard of event occurrence. In a sensitivity analysis, we will restrict the retention outcome to patients who have been in treatment for less than 12 months at the start of the study. The discrete-time multilevel model accommodates both time-invariant patient characteristics as well as time-varying patient and clinic characteristics as explanatory variables. Our focus will be on the time-varying intervention condition variable at the clinic-level, adjusting for patient characteristics such as demographics and prior treatment experience. Intervention effects will be visualized by plotting fitted hazard of dropout against month since treatment initiation separately for control and intervention observations. The multilevel discrete-time survival model can be used for other time to event outcomes such as time to all-cause mortality.

Aim 5. Assess clinic implementation outcomes and conduct qualitative interviews with OTP leadership, staff, and clients on attitudes, experiences and behaviors related to the intervention

We will examine OTP staff and leadership attitudes, experiences, and behaviors related to implementing the OTP intervention. We plan to collect diverse data: 1) implementation surveys, 2) semi-structured interviews, and 3) external facilitator notes, surveys, and checklists [114–116]. An online version of the Klein implementation survey [117, 118] will be administered to staff at 6-months post-intervention. Semi-structured interviews (n = 72) will be conducted with at least one clinical and one member of OTP executive leadership at each OTP at the end of the six-month intervention phase. An interview guide will be created to reflect the dimensions of the HEIF, including patient-provider encounter factors from the Kilbourne framework as well as implementation factors from the i-PARIHS model [63, 94, 95, 119]. In order to inform replication and dissemination of this intervention model, as well as contextual factors affecting sites’ ability to implement THD, the facilitator will complete logs after each interaction with a study site [114]. Client interviews (n = 40) from participating OTPs will explore experiences related to THD, notably among Black/African American and Latinx clients. We will include patients who are: 1) aged 18 or older; and 2) receiving take-home methadone for at least 30 days. An interview guide will be created to reflect the dimensions of the HEIF. We will use maximum variation purposeful sampling by recruiting clients from participating OTPs in urban and rural locations as well as areas with variation in population characteristics. Further, we will recruit clients who are on a variety of THD schedules and ensure that we have representation based on gender, race, and ethnicity. Clients will receive a $30 incentive to participate [120].

Analysis. For the survey results, staff responses will be aggregated and group means and standard deviations will be calculated. Multilevel modeling will be used to examine change over time in survey scores given the nested structure of the data (staff nested within clinics) [121]. The semi-structured interviews will be recorded and transcribed and then using directed content analysis following steps recommended for content analysis [85–87]. We will use Dedoose to manage, code, organize, and examine patterns in the data. Researchers will make field notes immediately after interviews. We will use a deductive approach such that a codebook with codes and operational definitions will be created prior to analysis using key elements from HEIF. We will use a continuous process of coding, categorizing, and reviewing the raw data to reflect on the analysis at various points and make revisions as needed (e.g., recoding data). Weekly meetings will be held to review coding decisions, discuss discrepancies, and check progress. We will use the first few transcripts to refine the coding scheme, and additional transcripts will be compared to previously coded transcripts to ensure the consistent assignment of codes. As emerging concepts are identified, we will adapt the existing coding structure. After the team has reviewed the coding structure and all initial transcripts have been reviewed in depth by the team, trained project staff will independently code all transcripts using the final coding scheme. Twenty percent of the transcripts will be double coded to assess inter-coder agreement. Any differences in coding will be discussed and resolved after discussion with the investigators. We also will create an analysis audit trail to document all analytic decisions.

Aim 6. Using mixed methods, explore variation in THD associated with race and ethnicity

Throughout quantitative analysis, we will examine by including additional terms in the mixed-effects logistic regression models and in the multilevel discrete-time survival model to incorporate interactions between the intervention effect and patient race/ethnicity. If significant interaction effects are detected, simple main effects of intervention will be estimated for each racial/ethnic group to understand how the intervention effect may vary and to identify any inequities in intervention impact. Throughout qualitative analysis, we will explore variation in THD experiences among Black/African American and Latinx clients and those clinics that serve more Black/African American and Latinx populations.

Dissemination

This study will comply with the NIH Public Access Policy, which ensures that the public has access to the published results of NIH funded research. It requires scientists to submit final peer-reviewed journal manuscripts that arise from NIH funds to the digital archive PubMed Central upon acceptance for publication. This trial has been registered with ClinicalTrials.gov database (#NCT05675735). The data generated in this study will be presented at national or international conferences and published in a timely fashion. All final peer-reviewed manuscripts that arise from this proposal will be submitted to the digital archive PubMed Central. Project data will be made available to the Data Ecosystem in a manner that is consistent with Data Use Agreements with the data owners.

Status and timeline

The recruitment of initial interviews to inform the design of the intervention has begun. The working groups are meeting regularly to design an intervention by the end of Year 1. Site recruitment will start by August 2023. All participating HCPs will complete training by Year 2. Data collection will be completed by August 2026. Data analysis and reporting of results will be completed by August 2027.

Conclusion

Currently, there is a national debate about balancing safety concerns over more flexible THD against the benefits of client retention and quality of life. Studies have not found increased overdoses or diversion under more flexible THD rules during the PHE; however, longer term studies are needed to understand best THD practices and outcomes. This study leverages large administrative datasets that can inform immediate policy questions of supporting flexible THD. The intervention targets improvement in treatment for OUD and tracks outcomes for healthcare, overdose, and mortality. It will develop and test actionable organizational interventions to promote better person-centered treatment.

Supporting information

S1 Checklist. SPIRIT 2013 checklist: Recommended items to address in a clinical trial protocol and related documents*.

(DOCX)

S1 File

(DOCX)

Data Availability

No datasets were generated or analysed during the current study. All relevant data from this study will be made available upon study completion.

Funding Statement

CN, KR, YB: R61 DA057683-01 National Institute of Health, National Institute on Drug Abuse: https://nida.nih.gov/ The funder did not and will not have a role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.