Inpatient adoption of medications for alcohol use disorder: A mixed-methods formative evaluation involving key stakeholders

Abstract

Background:

Although the inpatient setting presents an important opportunity for medications for alcohol use disorder (MAUD) adoption, this infrequently occurs. We aimed to develop a comprehensive understanding of barriers and facilitators of inpatient MAUD adoption.

Methods:

A convergent mixed-method study conducted from April to September 2018 of non-prescribing (registered nurse, pharmacist, and social work) and prescribing (physician or advanced practice provider hospitalist, general internist, and psychiatrist) professionals at a large urban academic medical center. Survey assessed organizational readiness to adopt MAUD and focus groups guided by the Consolidated Framework for Implementation Research (CFIR) analyzed using directed content analysis.

Results:

Fifty-seven participants completed surveys and one of seven focus groups. Health professionals perceived clinical evidence (mean 4.0, 95% confidence interval [CI]: 3.9, 4.2) as supportive and patient preferences (mean 3.4, 95% CI: 3.2, 3.6) and availability of resources (mean 3.1, 95% CI: 2.8, 3.3) as less supportive of MAUD adoption. Stakeholders identified barriers across CFIR constructs; 1) Intervention characteristics: limited knowledge of MAUD effectiveness and concerns about side effects, 2) Outer setting: perceived patient vulnerability to care interruptions and a lack of external incentives, 3) Inner setting: a lack of organizational prioritization, and 4) Characteristics of individuals: stigma of people with AUD. Facilitators included: 1) Intervention characteristics: adaptation of workflows and 2) Characteristics of individuals: harm reduction as treatment goal.

Conclusions:

This study identified multiple intersecting barriers and facilitators of inpatient MAUD adoption. Implementation interventions should prioritize strategies that increase health professional knowledge of MAUD and organizational prioritization of treating AUD.

1. INTRODUCTION

Alcohol use disorder (AUD) is a leading cause of morbidity and mortality globally (Griswold et al., 2018). The prevalence of AUD among US adults increased from 20.2 to 29.9 million between 2001 and 2013 with over 88,000 alcohol-related deaths occur annually (Centers for Disease and Prevention, 2004; Grant et al., 2017). Despite the rising morbidity, less than 15% of adults with AUD report receiving indicated treatment (MAUD or other treatment modality) (Cohen et al., 2007). Notably, medications for AUD (MAUD) include Food and Drug Administration approved acamprosate, naltrexone (oral and intramuscular formulations), and disulfiram as well as off-label gabapentin and topiramate. Use of these medications is widely recommended by numerous clinical organizations and guidelines (Agency for Healthcare Research and Quality, 2016; Anton et al., 2006; Donoghue et al., 2015; Jonas et al., 2014; Reus et al., 2017; SAMHSA, 2015;). However, less than 9% of patients with AUD receive MAUD across combined inpatient and outpatient settings (Del Re et al., 2013; Harris et al., 2010; Iheanacho et al., 2013; Owens et al., 2018; Thomas et al., 2016; Williams et al., 2017), and evidence suggests lower adoption within emergency room and inpatient settings. Among patients with AUD receiving care at a large urban academic medical system without engagement with primary care services, less than 2% received a MAUD prescription (Joudrey et al., 2019).

An important opportunity for promoting MAUD adoption is hospitalization. First, alcohol-related hospitalizations are the leading cause of substance use-related hospitalizations within the United States (Owens, 2019) and patients may not otherwise receive routine care (Joudrey et al., 2019). Second, there is the potential for repeated interaction with multidisciplinary teams of health-professionals who can engage patients about MAUD. Third, patients may be more motivated to make a change in their alcohol use in the context of illness (Velez et al., 2017). Fourth, MAUD delivery may serve to prevent future readmissions and reduce the utilization of health care services among a high-risk population (Wei et al., 2015). Fifth, emerging data on hospital-based initiatives with addiction consult services indicate that addiction treatment translates into reduced substance use post-discharge (Englander et al., 2019; Wakeman et al., 2017), lending support for this approach to care.

Data to inform efforts to optimize MAUD adoption in the inpatient setting are currently lacking. Studies examining barriers and facilitators of MAUD adoption thus far have focused on outpatient settings, such as primary care, and it is unclear if previously identified barriers such as limited knowledge of MAUD and stigma of AUD (unfavorable attitudes, beliefs, and policies directed toward people with AUD) (Room, 2005) extend to the inpatient setting and what novel inpatient barriers and facilitators may exist (Hagedorn et al., 2016; Harris et al., 2017; Harris et al., 2016; Robinson et al., 2013; Storholm et al., 2017). Additionally, among existing studies of implementation interventions to improve MAUD adoption, consensus on the optimal implementation strategy is missing, suggesting a further need for formative research on the antecedents of implementation (Williams et al., 2019). To address this critical gap and inform future implementation efforts, we sought to develop a comprehensive understanding of barriers and facilitators of MAUD adoption among hospital-based health professionals who may routinely participate in care for patients with AUD.

2. METHODS

2.1. Study overview and sample

Guided by the Consolidated Criteria for Reporting Qualitative Research (COREQ) and Good Reporting of A Mixed Methods Study (GRAMMS) guidelines (O'cathain et al., 2008; Tong et al., 2007), we conducted a mixed-methods formative evaluation at Yale New Haven Hospital (YNHH), an academic medical center and integrated health care delivery system in New Haven, CT with nearly 80,000 inpatient visits and nearly 1.4 million outpatient visits annually. At the time of this study, no implementation interventions existed to promote the adoption MAUD at YNNH. We purposively sampled non-prescribing (registered nurse, pharmacist, and social work) and prescribing (physician or advanced practice provider hospitalist, general internist, and psychiatrist) professionals to contrast both groups’ roles in MAUD adoption on the general medicine inpatient service. We collected quantitative and qualitative data concurrently from April to September of 2018, consistent with a convergent approach, and integrated data during analysis (Fetters et al., 2013). We grounded our work in the Consolidated Framework for Implementation Research (Table 1) (Damschroder et al., 2009). The CFIR is a commony cited implementation framework which organizes the factors associated with the successful adoption of evidence-based practice. The CFIR contains five major constructs and has been used to support a consistent and systematic approach to evidence-based practice adoption across a variety of clinical settings (Kirk et al., 2015). The Institutional Review Board of Yale University approved this study and participants received lunch but no other compensation.

Table 1:

Consolidated Framework for Implementation Research (CFIR) constructs included during directed content analysis

Construct*	Topic
Intervention Characteristics: medications for alcohol use disorder	Intervention source Evidence strength and quality Relative advantage
	Adaptability
	Trialability
	Complexity
	Design quality and packaging Cost
Outer Setting: policy and community environment outside of the hospital	Patient needs and resources Cosmopolitanism Peer pressure
	External policies and incentives
Inner Setting: general medicine hospital setting	Structural characteristics Networks and communications Culture
	Implementation Climate
	Readiness for implementation
Characteristics of individuals: health professionals	Knowledge and beliefs Self-efficacy Individual stage of change
	Identification with organization
	Other personal attributes

^*The fifth construct, process, was not included given this is a formative evaluation.

2.2. Data collection

2.2.1. Quantitative data

We collected demographic data including age, gender (male, female, other), race/ethnicity (Hispanic/Latino of any race, White non-Hispanic, Black non-Hispanic, and other/unknown), profession, and years in current profession. We collected perspectives on organizational readiness to prescribe MAUD, assessed using an adapted version of the Organizational Readiness to Change Assessment (ORCA) (Appendix 1²) (Helfrich et al., 2009). The ORCA is a validated instrument to assess an organization’s readiness to adopt an evidence-based practice (Helfrich et al., 2009). Participants completed the first two scales of the ORCA: 1) strength and nature of evidence as perceived by stakeholders supporting change in evidence-based practice and 2) quality of the context or environment in which research is implemented. The evidence scale consists of four subscales and the context scale consists of six subscales. The first evidence subscale examines the discrepancy between the respondent's opinion of the strength of evidence and that of peers with scores approaching zero suggesting no discrepancy. All other subscales (evidence and context) are scored on a one to five Likert scale with higher scores suggesting increasing organizational readiness for evidence-based practice (Hagedorn and Heideman, 2010).

2.2.2. Qualitative data

Our research team included general internists, addiction medicine specialists, and one psychiatrist, with health services research training (Barry et al., 2009; Oldfield et al., 2019). We conducted focus groups until thematic saturation (i.e., no new ideas emerged across the entire sample) (Curry et al., 2009). In collaboration with leadership of each professional group, focus groups were conducted during pre-existing meeting times (i.e., noon conferences or shift changes at YNHH), with groups specific to profession to ensure each profession was represented within our sample and to minimize differences in power (Willis et al., 2009). Participants were not recruited in advance of each pre-existing profession meeting and participants did not know the topic of the meeting prior to attending. Verbal consent was obtained. All research team members participated in the development of the focus group guide (Appendix 2³) with lead questions and probing questions organized by the first four CFIR constructs (Damschroder et al., 2009) and refined iteratively. The first focus group was facilitated by PJ and BO and all others were facilitated by PJ. Focus groups were audio recorded and professionally transcribed.

2.3. Data analysis

2.3.1. Quantitative data

We used descriptive statistics for participant characteristics. We calculated ORCA evidence and context subscale scores by dividing the total score by the number of sub-scale items. Mean subscale scores were then calculated among all participants. Individuals with a missing ORCA item within a given subscale were excluded from mean scores (Kang, 2013). ORCA subscale scores were then stratified by prescribing status (prescribing and non-prescribing) and compared using a two-sample t-test (two-sided with an alpha .05). We completed all quantitative analysis using Stata 15 (StataCorp, College Station, Tx).

2.3.2. Qualitative data

Our research team held seven meetings, each one to two hours duration, during data collection and analysis to discuss focus group content, the focus group guide, and new topics of interest. We analyzed transcripts using directed content analysis (Hsieh and Shannon, 2005) grounded by CFIR. The goal of directed content analysis is to apply an existing theoretical framework to a new setting or population. To develop a preliminary code list, team members reviewed three transcripts applying CFIR constructs to segments while noting additional ideas and relationships. After initial coding, we held a consensus meeting where additional sub-codes within the CFIR constructs were named, and a code list was refined. We repeated this process during meetings over a three-month period until a final code book was agreed upon. PJ then coded all transcripts and BO reviewed the coding to ensure agreement. Upon organization of segment text by code, we used an iterative inductive approach to develop emerging themes. Upon completion of our analysis, we shared preliminary results with participants at noon conferences and team huddles to elicit feedback. To organize data and facilitate retrieval, we used Dedoose (2018 version, SocioCultural Research Consultants, Los Angeles, CA) software.

2.4. Mixed methods integration

We used a convergent approach (Fetters et al., 2013) to quantitative and qualitative integration where early themes from our focus group analysis were compared with results from our ORCA analysis (Damschroder et al., 2017). Areas of convergence and divergence were identified and used to inform finalization of major themes.

3. RESULTS

3.1. Participant characteristics

A total of 57 participants completed surveys and participated in one of seven focus groups with at least five participants per each profession group (Table 2). All healthcare professionals approached for participation participated in our study.

Table 2:

Demographic survey and focus group participant characteristics

Characteristic	Participants (n=57)
Age, median (IQR)	34 (29, 44)
Race/ethnicity, n (%)
Hispanic	2 (4)
Non-Hispanic Black	7 (12)
Non-Hispanic White	38 (67)
Asian	10 (18)
Female gender, n (%)	43 (75)
Years in current profession, median (IQR)	4 (2, 12)
Profession (non-prescribing*), n (%)	27 (47)
Registered nurse	8 (14)
Pharmacist	9 (16)
Social worker	10 (18)
Profession (prescribing*), n (%)	30 (53)
Nurse practitioner	6 (11)
Physician assistant	8 (14)
Physician hospitalist	5 (9)
Physician general internist	5 (9)
Physician psychiatrist	6 (11)

^*Based on whether the health professional role allows prescribing medications for alcohol use disorder (MAUD)

3.2. Organizational readiness to adopt MAUD

Among all participants, the response rate for each ORCA subscale was 84% or greater. Within the evidence scale (Table 3), there was no discrepancy about the strength of the evidence between respondents and peers. Within the evidence scale, scores were highest for the research evidence subscale and lowest for patient preferences subscale. Within the context scale, scores were highest for staff culture and lowest for availability of resources to support MAUD adoption. Comparing subscale results by prescribing status, prescribing professions reported lower availability of resources to support the adoption of MAUD relative to non-prescribing professions (mean difference 0.5, 95% CI: 0.01, 0.91). There was no difference by prescribing status for all other subscales.

Table 3:

Organizational Readiness for Change Assessment (ORCA) survey results among general medicine hospital-based health professionals (n =57)

Scales and Subscales*	n‡	All providers (n=57), mean (95% CI)	n‡	Non- prescribers (n=27), mean (95% CI)	n‡	Prescriber (n=30), mean (95% CI)	p – value§
Evidence
Discordance about evidence†	48	0.1 (−0.1, 0.4)	25	0.0 (−0.4, 0.4)	23	0.2 (0.0,0.5)	0.46
Research evidence	48	4.0 (3.9, 4.2)	25	4.0 (3.8, 4.2)	23	4.1 (3.9, 4.3)	0.57
Clinical experience	50	3.5 (3.2, 3.7)	26	3.5 (3.1, 3.8)	24	3.5 (3.1, 3.9)	0.99
Patient preferences	54	3.4 (3.2, 3.6)	26	3.5 (3.2, 3.9)	28	3.3 (2.9, 3.6)	0.29
Context
Leadership culture	56	3.7 (3.5, 3.9)	27	3.7 (3.4, 3.9)	29	3.7 (3.4, 4.0)	0.79
Staff culture	56	4.0 (3.8, 4.1)	27	3.9 (3.7, 4.1)	29	4.1 (3.8, 4.3)	0.26
Leadership practices	56	3.8 (3.6, 4.0)	27	4.0 (3.8, 4.2)	29	3.7 (3.4, 4.0)	0.15
Measurement	55	3.7 (3.5, 4.0)	26	3.8 (3.5, 4.1)	29	3.7 (3.3, 4.0)	0.53
Readiness for change	56	3.8 (3.6, 3.9)	27	3.7 (3.5, 4.0)	29	3.8 (3.6, 4.0)	0.50
Availability of resources	53	3.1 (2.8, 3.3)	27	3.3 (3.0, 3.6)	26	2.8 (2.5, 3.2)	0.048

^*Third scale, facilitation, was not included given this is a pre-implementation study.

^†Subscale examines the perceived discrepancy in the strength of evidence between peers with scores approaching zero suggesting no discrepancy. All other subscales are scored on a 1 to 5 Likert scale with higher scores suggesting increasing organizational readiness for adoption.

^‡Number of participants who completed all items of a given subscale.

^§Mean non-prescribing vs prescribing subscale scores using a two sample ttest, two-sided with alpha of .05.

3.3. Barriers and facilitators to MAUD

We identified barriers and facilitators (Table 4) to inpatient MAUD adoption among key stakeholders. We present themes and representative verbatim quotes organized by relevant CFIR constructs (bold heading) and subconstructs (italic heading).

Table 4:

Identified barriers and facilitators of MAUD adoption within the hospital and implementation strategies for further investigation

Construct	Barrier or facilitator	Implementation strategy‡
Intervention Characteristics: medications for alcohol use disorder	Administrative complexity Anticipated low patient engagement†	Develop and distribute educational materials Conduct educational outreach visits (academic detailing) Promote adaptability by reducing prescribing requirements
Outer Setting: policy and community environment outside of the hospital	Perceived patient vulnerability to care interruptions Lack of AUD-related incentives	Develop and implement tools of quality monitoring Linkage to care interventions
Inner Setting: general medicine hospital setting	Lack of organizational prioritization of AUD Limited time to deliver direct patient care† Adaptation of exiting workflows Missed opportunities for care coordination	Change record systems Audit and feedback Identify and prepare champions Involve executive boards Use advisory board and workgroups Promote network weaving across disciplines
Characteristics of individuals: health professionals	Limited knowledge of MAUD AUD exceptionalism among other chronic diseases Harm reduction as treatment goal	Develop and distribute educational materials Conduct educational outreach visits (academic detailing)

^†Converging ORCA and focus group results

^‡Recommended implementation strategies for further investigation

3.4. Intervention characteristics: Medications for alcohol use disorder

Prescribers and non-prescribers expressed limited or inaccurate knowledge about MAUD. Among prescribers this fostered a perception that these medications were complex and unsafe to initiate inpatient. This contrasted with perspectives among non-prescribers, however, who viewed hospitalization as an important opportunity for adoption.

3.4.2. Complexity

Intervention complexity was increased by perceived administrative barriers, such as hospital policy impacting naltrexone injections. At the time of the study, YNHH did not have any special requirements for the initiation of naltrexone injections, but more restrictive prior policies created confusion.

…to give a Vivitrol [naltrexone] injection before a patient’s discharge, it has to be approved by an addiction specialist. I think it’s a giant barrier. (Physician general internist)

3.4.3. Adaptability and relative advantage

Prescribers overestimated the risks of inpatient MAUD adoption creating doubts about the adaptability of MAUD to the inpatient setting. These adaptability concerns were heightened given expectations of poor outpatient follow-up upon discharge and stigma of people with AUD. Together, this diminished their enthusiasm for inpatient MAUD adoption relative to outpatient adoption and they favored that MAUD initiation occur in the outpatient setting. For example, one psychiatrist expressed an unwarranted concern that a patient might overdose on MAUD despite the safety of these medications:

When you don’t know when they’re going to see someone to continue the medication, then you’re very hesitant to start something that they’re not going to continue that they could potentially overdose on. Unfortunately, there’s also a lot of defensive medicine that we practice, that we don’t want bottles with someone who’s not responsible. (Physician psychiatrist)

In contrast, non-prescribers recognized hospitalization as an ideal time for recurrent conversations and patient engagement regarding MAUD relative to outpatient settings:

Starting it here in the hospital, being surrounded by people who can educate them, support them, help them, I think is more of an opportunity for a success on discharge and further on. (Registered nurse)

3.5. Outer setting: Community environment and policies outside of the hospital

Anticipated interruptions in post-hospital discharge care due to social determinants and a lack of attention to AUD-related quality measures reduced MAUD adoption.

3.5.1. Patient needs and resources

Despite being in a Medicaid-expansion state, health professionals perceived patients with AUD as vulnerable to interruptions in care due to poverty, co-morbid mental health disorders, and a lack of provider continuity. Health professional anticipation of care interruptions following discharge from the hospital due to these unaddressed social needs reduced MAUD adoption.

But the patient could come from a very negative social background. Like, the patient is homeless, for example. And there’s really no way for the patient to get medications, so there’s really no incentive for me as a professional to go in and do a medication reconciliation…I know that the minute the patient actually steps out the door, they’re not going to be compliant. (Pharmacist)

3.5.2. External policies and incentives

Health professionals perceived an absence of AUD-related quality measures for the inpatient setting (measures used to monitor quality across health care systems) and contrasted this with the importance placed on quality measures for other chronic diseases, such as heart failure. Health professionals understood this difference in the prioritization of AUD related measures relative to measures associated with other chronic diseases in the outer setting as unfavorable treatment of people with AUD (stigma).

When you discharge someone with heart failure, there’s going to be a bunch of quality measures. Either they go home on an ACE inhibitor or a beta blocker, they can follow up with you in X amount of time. There’s nothing like that for medications to prevent alcohol relapse. (Physician general internist)

3.6. Inner setting: Inpatient setting

Low organizational prioritization of AUD, limited time for direct patient care among prescribers relative to non-prescribers, and missed opportunities to coordinate care were perceived to reduce MAUD adoption.

3.6.1. Implementation climate

Health professionals perceived AUD as a low organizational priority relative to other chronic diseases given the limited availability of support services for patients with AUD.

With heart failure, if it’s identified in the hospital, you get the heart failure clinic coming in and saying, “Oh my goodness, you have a new onset heart failure. This is what we’re going to do for you, and this is how often you’re going to see us.” Somebody who’s coming in with an ETOH fall at home, broke their nose, who’s coming into their room? (Social worker)

I know we don’t prioritize [patients with AUD], because some of our more high-risk patients we do discharge med recs on, being heart failure, HIV patients. I know alcohol use, we wouldn’t usually do a discharge med rec on that patient population. (Pharmacist)

3.6.2. Readiness for implementation

The perceptions of the resource of time delivering direct patient care to patients with AUD varied across prescribing and non-prescribing members of the hospital team.

We have so many people to see and so many things that come up over the course of the day. It takes time to establish a relationship with somebody, to talk about their alcohol use disorder. We don’t have the resource of time to sit down with somebody and engage in that interview, like motivational interviewing. Within the structure of our hospital and our team and the other patients, the other strictly medical patients, it’s really hard to sit down with somebody and dig that deep. (Advanced practice provider, hospitalist)

You know you spend 12 hours and then maybe, like, multiple shifts with them, you get to know them. They might mention things when it’s just you and them and maybe they’ve had breakfast and they’re cleaned up and they feel, like, “Yeah, let’s chat about that.” You know, we can- we can kind of figure out when we can seize those opportunities. (Registered nurse)

Additionally, non-prescribers recognized, if they had greater knowledge, their existing workflows could be adapted to support MAUD adoption.

I think that if we did know that, it would provide us with confidence to educate our patients like we do with smoking. (Registered nurse)

3.6.3. Networks and communication

Given multidisciplinary team-based care in the hospital, health professionals described missed opportunities to coordinate MAUD adoption. For example, one prescriber was clear in their lack of active engagement in standing multidisciplinary meetings to address treatment needs of patients with AUD:

We have a dedicated social worker that deals with alcohol. Those people spend a lot of time with those patients. I certainly don’t go to those conferences with them, so I never really know what’s being discussed. (Physician hospitalist)

Despite being acknowledged as key team members in facilitating treatment of patients with AUD, one non-prescriber described how she did not feel empowered to raise the issue of MAUD given her training and role:

I don’t ever bring it up because I try not to ever touch meds because I feel like it’s not my role. I feel like sometimes the medical team does bring it up. Not even close to as much as I would like, but fragments of the time. (Social worker)

3.7. Characteristics of individuals: Health professionals

Health professionals believed AUD was different from other chronic diseases and that MAUD adoption was contingent on a goal of abstinence.

3.7.1. Knowledge and beliefs

Prescribers and non-prescribers acknowledged having limited clinical training on MAUD.

If I think back in school, how many lectures did I get about substance [use]? It was probably one hour, maybe. I think a lot of us would need more education about this stuff. (Advanced practice provider hospitalist)

Non-prescribers and prescribers alike viewed AUD as exceptional to other chronic diseases, such as diabetes mellitus, managed within the hospital consistent with stigma of AUD. One nurse described how AUD involves a unique behavioral component to justify differences in care.

…as far as discharging them and making sure that they have insulin versus making sure that they don’t drink again, I think that insulin’s a lot easier. You know, to make sure that they go home with insulin, it doesn’t have all the social stuff attached to it — the social stuff that’s attached to a diagnosis of [AUD] it is a lot harder for them to walk out of here with [MAUD] than insulin for diabetes. [Registered nurse)

Furthermore, prescribers and non-prescribers viewed MAUD adoption as contingent upon a patient’s commitment to abstinence from alcohol use. One prescriber described:

I will say I select certain medications, sometimes I won’t prescribe a medicine if a patient is really not ready to quit drinking. I will select certain medications depending on what situation they have at home. [Physician general internist)

This was consistent with one pharmacist’s observations:

I think maybe a lot of providers in the inpatient side aren’t familiar with harm reduction strategies as well, versus these medications being used for complete abstinence. (Pharmacist)

4. DISCUSSION

To our knowledge, this study is the first to identify barriers and facilitators of MAUD adoption within the hospital and notably includes the perspectives of both prescribing and non-prescribing health professionals. We identified several potentially addressable barriers to inpatient MAUD, including a lack of knowledge, concern about discharge follow-up, low organizational prioritization, and stigma of people with AUD. We also identified important facilitators to MAUD adoption: the perceived potential to leverage existing workflows and enthusiasm to promote evidence-based AUD care among non-prescribers.

Our findings are consistent with prior research in other treatment settings and identifies which barriers and facilitators extend to the inpatient setting. For example, limited knowledge has also been identified as a barrier to MAUD adoption within primary care and addiction specialty settings (Finlay et al., 2017; Harris et al., 2013; Storholm et al., 2017; Williams et al., 2018). Anticipated low patient engagement with care and a lack of time for direct alcohol-related care were also identified barriers within the addiction specialty setting (Harris et al., 2013). Despite growing attention to the importance of addressing substance use, our results also demonstrate that stigma of people with AUD continues to be an important problem. Consistent with prior literature on stigma of substance use disorders (Kulesza et al., 2013; McCradden et al., 2019; Room, 2005), evidence of stigma of AUD extended across societal (outer setting), organizational (inner setting), and individual (health professional) constructs. There was a perceived absence of AUD-related quality measures relative to other chronic diseases within the outer setting and a low organizational prioritization of AUD in the inner setting. The low prioritization of AUD within the outer and inner settings reinforced and confirmed health professionals own attitudes of AUD exceptionalism, resulting in differences in care for AUD relative to other chronic diseases. Quality measures for MAUD exist (Harris et al., 2016; American Society of Addiction Medicine, 2014), but they did not influence MAUD adoption among the hospital-based health professionals sampled given their limited awareness of these measures.

Our study of barriers and facilitators within the inpatient setting also contrasts with prior research in other treatment settings. Previous research within primary care and residential treatment settings found an organizational preference for psychosocial treatment or 12 step programs were a barrier to MAUD adoption (Finlay et al., 2017; Williams et al., 2O18). Such a preference did not emerge as a barrier among our hospital-based sample; however, professionals who framed the goal of treatment in terms of abstinence were less favorable towards MAUD. Perceived patient vulnerability to interruptions in care and the low prioritization of AUD quality measures were unique external barriers to MAUD adoption in the hospital and were not previously identified in other settings (Finlay et al., 2017).

Our identification of inpatient barriers and facilitators suggests important opportunities for tailoring future MAUD implementation research within the inpatient setting and suggests addressing inpatient stigma of AUD will require incorporation of multiple implementation strategies spanning the CFIR constructs. Informed by previous MAUD implementation interventions in other treatment settings, table 4 presents common implementation strategies paired with specific barriers and facilitators across CFIR domains for future investigation. A previous academic detailing intervention (non-commercial face-to-face educational outreach visits) within the Veterans Affairs Healthcare System improved MAUD adoption and academic detailing within the inpatient setting may address the lack of knowledge and stigma of AUD among inpatient healthcare professionals (Harris et al., 2017; Soumerai and Avorn, 1990). The promotion of network weaving (information sharing and collaborative problem solving within organizational teams) may facilitate empowerment of non-prescribing inpatient team members to utilize existing care pathways for MAUD adoption. Strategies such as identifying champions and auditing MAUD adoption rates and providing feedback to inpatient professionals (Powell et al., 2015) are needed to increase organizational prioritization of MAUD in the inner setting. Likage to care interventions may reduce concerns about patient vulnerability to care interruptions following discharge in the outer setting (Priest and McCarty, 2019). Finally, our study suggests current MAUD-related quality measures are not impacting health professional behavior, and greater promotion of MAUD-related measures by professional and government agencies should be investigated.

Our study has several limitations. First, it does not capture patient perspectives. However, our findings complement prior research that found hospitalized patients with substance use disorders were unfamiliar with MAUD and perceived that health professionals lacked knowledge about SUDs and viewed the disorder as a moral failing (Velez et al., 2017). Second, our sample is limited to a single urban academic hospital center, which may impact generalizability to other settings, particularly rural settings, non-academic hospitals, and hospital services outside of general medicine.

5. CONCLUSION

In conclusion, we identified multiple levels of intersecting barriers and facilitators of MAUD adoption within the hospital. Future implementation interventions should utilize academic detailing, identify organizational champions, and improve post discharge linkage to care to promote inpatient MAUD adoption. Further, global policy changes, including a focus on MAUD quality of care tied to financial incentives, are needed to improve MAUD adoption.

HIGHLIGHTS.

• Inpatient prescribing of medications for alcohol use disorder infrequently occurs

• Convergent mixed-method study of barriers and facilitators of medication adoption

• Barriers include limited knowledge of effectiveness and lack of prioritization

• Facilitators include adaptation of workflows and harm reduction as treatment goal

ABBREVIATIONS

MAUD

medications for alcohol use disorder

AUD

alcohol use disorder

CFIR

consolidated framework for implementation research

ORCA

organizational readiness for change assessment