Overdose is a leading cause of injury-related death in the US,1 with 107 941 such deaths occurring in 20222 and annual deaths due to alcohol exceeding 140 000 from 2015 to 2019.3 The more than 46.3 million people in the US with a past-year substance use disorder4 and a nationwide economic impact of alcohol misuse and illicit drug use that tops $442 billion5 further evidences the magnitude of this crisis.
A variety of safe and effective evidence-based practices (EBPs) to identify, reduce the morbidity and mortality of, and treat substance use disorders exist. Examples include screening, brief intervention, and referral to treatment,6–10 as well as behavioral therapies and pharmacotherapies for nicotine, alcohol, and opioid use disorders.11–13 Furthermore, harm reduction approaches (eg, naloxone training and coprescribing, drug checking and testing, and syringe service programs) offer significant individual and public health benefits for people who use drugs and for those who do not have abstinence-based treatment goals.14–16
Clinician adoption of EBPs is necessary; however, screening for substance use disorders remains low,7 creating missed opportunities to intervene in harmful substance use or recognize and discuss potential progression to a severe disorder. Treatment capacity is inadequate to meet demand,17 with only 6.3% of people with a past-year substance use disorder receiving treatment in the US in 2021.4 Our goal is to summarize published data on physician-described barriers to adoption of EBPs for addiction in clinical practice and recommend actions to address them.
Data Sources and Searches
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines. The search strategy was developed iteratively with a National Library of Medicine informationist specializing in systematic reviews. We applied this strategy on October 4, 2021, to PubMed, Embase, and Scopus and on October 5, 2021, to medRxiv and SSRN Medical Research Network. In addition, a gray literature search of relevant government and nongovernment websites was conducted on October 5, 2021. We found no previous similar systematic reviews. The systematic review protocol was registered in PROSPERO (CRD42022286208) and accepted on January 14, 2022.
A 12-person team used Covidence to apply exclusion criteria first to the title and abstract of each study then to the full text of studies not already excluded. Two people (L.N., M.C., L.F., J.P., C.S., and S.W.) reviewed each study in both rounds. Discordant opinions were resolved by a third reviewer (M.C. and W.C.). To be included, the study had to present data on: (1) physicians at any practice level; (2) any substance use intervention(s) (Box); and (3) physician reasons for reluctance to intervene in addiction. Studies not in English, letters, editorials, narrative reviews, and commentaries were excluded. Data collection on reasons for reluctance were systemized using the theoretical domains framework (TDF),18,19 a comprehensive approach for identifying behavioral determinants and for assessing implementation problems (eg, clinicians’ behavior) to inform intervention development. The team created a data extraction template with 10 reluctance reason categories (Box). We did not formally assess risk of bias in included studies because few used experimental or controlled study designs. Due to patterns observed during data extraction, the team approved the ad hoc collection of data on factors (eg, using a theoretical framework, obtaining target audience input in survey design, and piloting surveys) that could affect the internal validity of individual studies or precision of results. We conducted a limited exploration of facilitators because we observed that many included studies provided at least some data on possible facilitators of intervention in addiction.
Definitions of Intervention Type and Reluctance Reasons
Intervention type and definition
-
Harm reduction: syringe services, overdose prevention, naloxone, or drug user health.
-
Screening and assessment: screening, assessment of positive screening, or diagnosis.
-
Treatment: brief intervention, medication management, or behavioral services.
-
Recovery support: care coordination, care integration, or relapse prevention.
Reason and definitiona
-
Knowledge: beliefs about having the necessary knowledge, awareness, or understanding, including knowledge of condition or scientific rationale, procedural knowledge, or knowledge of task environment.
-
Institutional environment: beliefs about support from institution or employer, including material resources, organizational culture, competing demands.
-
Skills: beliefs about having the necessary skills, ability, or proficiency to deliver the intervention.
-
Cognitive capacity: beliefs about the cognitive capacity to manage a level of expected complexity of care, possibly related to cognitive overload and mental fatigue.
-
Expectation of benefit: beliefs about the likelihood of the patient benefiting or the course of the disease being altered due to the intervention.
-
Social influences: beliefs about public or community acceptance or support for the intervention, including willingness to allocate or develop needed resources.
-
Emotion: feelings of fear, dislike, worry, negative judgement, worthiness of patient population.
-
Relationship: concern about harming or losing the patient-physician relationship by causing offense, provoking avoidance, or other negative consequence.
-
Reinforcement: beliefs about the adequacy of reimbursement, professional rewards, and other positive reinforcement.
-
Professional role/identity: beliefs about professional role, boundaries, and group identity, excluding the intervention.
a Reasons are derived from the theoretical domains framework, a comprehensive approach for identifying behavioral determinants and assessing implementation problems (eg, clinicians’ behavior) to inform intervention development.
We conducted a series of quantitative analyses using SPSS, version 27 (IBM). Analyses were selected based on their purpose; independent variable; dependent variable; and statistical requirements, including measurement levels. We examined reasons for reluctance by specialty, intervention, drug type, and year and common combinations of reasons for reluctance using bivariate analysis and cross-tabulation. We conducted a regression analysis of reasons for reluctance by year. Statistical significance was considered a 2-sided P value less than .05. The exploratory analyses of ad hoc study quality data were not part of the planned analysis and are descriptive only. We used Atlas.ti version 24 (Atlas.ti) to conduct thematic analysis to examine facilitators using the following themes: knowledge and skills, intrapersonal and interpersonal factors, infrastructure, and regulation reform.
Our search yielded 9308 studies published between January 1, 1960, and October 5, 2021, with 1280 remaining after removal of duplicates and 552 assessed for eligibility (eFigure 1 in Supplement 1). Of 283 studies20–302 included (eTable 1 in Supplement 1), 97.30% were published in 2000 or later (Table 1). The number of studies increased over time. For example, 4 studies89,156,184,236were published in 2000 and 2133,48,49,66,68,75,77,79,93,107,108,113,139,142,148,240,251,255,302,306,313 in 2021, with a high of 31 8,27,47,50,52,54,69,74,92,100,114,121,146,147,161,165,174,182,191,193,199,204,206,209,221,247,263,270,275,287,300 in 2020 (eTable 2, eTable 3, eTable 4, eTable 5, and eFigure 2 in Supplement 1). Together, the included studies describe the views of 66 732 physicians who largely practiced general practice, internal medicine, or family medicine primarily in an office setting in the US. Most studies reported survey-based research results. Of the 4 general categories of addiction interventions (Table 2), treatment was most often addressed, followed by screening and assessment, with harm reduction and recovery support least discussed. Some studies addressed more than 1 intervention. Alcohol (86 studies20,21,23,25,26,29,31,34,36,38,41,44,51,53,54,57,59,60,62,69–72,81,82,86,88,89,94,95,103,105,111,113,117,119,123–127,131,132,138,141,150,153,155,158,160,162,164,168,170,171,173,176,191–193,196–201,204,205,210,219,235,237,248,250,254,256,258,271,281,283,285,291,294,296,299,300), nicotine (30 studies28,40,48,49,52,61,73,85,97,109,118,129,134,140,142,149,179,188,190,212,218,223,231,249,252,265,270,286,288,298), and opioids (104 studies30,32,33,35,37,42,46,47,50,55,56,58,64,66,74–80,83,84,87,90–92,98–100,104,106–108,110,112,114,115,121,122,130,133,135,137,139,143,144,146–148,151,152,154,156,163,165,167,172,174,180,182,184,186,189,202,203,206,207,213–216,221,222,225–228,238–240,242–245,247,251,253,255,257,259,262,269,272,275,277,280,282,284,287,290,292,293,302) were most often studied alone. Among studies reporting on multiple drugs (44 studies22,39,43,45,63,65,67,68,93,96,101,102,116,120,136,145,166,181,183,185,194,195,208,209,217,220,230,232–234,241,246,260,263,264,267,268,273,274,278,279,289,295,297), alcohol was included most often (38 studies45,63,65,67,68,93,96,101,102,116,120,136,145,166,181,183,194,195,208,209,217,230,232–234,241,246,260,264,267,268,273,274,278,279,289,295,297). Other substances were often reported as “other” or merely “drugs.” Cross-tabulations of each reason for reluctance with each of the most common specialties, interventions, and drugs produced no significant results; consequently, no P values are reported (Table 2). While this systematic review is of physician reluctance, 110 studies20,23–25,28,30,31,33,34,39,42,44,47,48,50,52,54,57,59,63,64,66–70,87,88,90,92,93,95,99,101,103–107,109,111–113,116,120,122,123,126,129,134,136,138,139,143,146,147,151,156,157,159,162,166,167,169,173,174,177,178,183,186,189,190,192,194,195,199–201,203,205,206,209,211,217,221,225,229,235,236,243–245,251,257,260,261,266,269,270,275,277,280,283,286,287,290,291,297,299,302 mentioned possible facilitators of physician engagement.
Most studies did not gather or report data on all reasons. When queried, institutional environment (173 of 213 articles [81.2%]20,22,25–27,30–33,35,37,38,40–44,46,47,49–51,54–64,66,68,74–78,80,82–84,86,87,89–93,95,97,99,100,104,106–110,112–114,116,117,121–124,126,127,129,134–139,143,144,146–148,150,151,153–155,157–159,161–165,167,169–176,179,180,182,183,185,186,189,192,195,198,199,201–204,206,207,209,211,216–221,223,226,228–230,232–234,236,238,239,241–243,245,247,251,252,257–261,263–265,268,269,271,272,275,277,280,284,287,290,291,293,295,299,301,302) was the most common reason, followed by lack of skill (170 of 230 articles [73.9%]20–22,24–33,35,37–39,47–49,51,53–55,58,59,61,63–68,75,76,78,80–82,84,85,88,89,91–93,95,97–100,102–107,109–114,116–121,123–125,130–132,134,136,138,139,142,143,145,147,149,150,152,154,159–161,167,168,172–174,176,178,180,182,183,186,188,190,191,193,194,197–202,204,206–211,213,214,216,218–221,224–226,229,231,233,235,236,238,241,242,246,247,249,256,259,264–266,268,269,271,273,274,276–279,281–283,285–287,290–295,297,298,301,302), cognitive capacity (136 of 185 articles [73.5%]22,25,26,30,32,34,37,40,41,47–49,52,55,58–61,63–66,68,69,71,74,75,77,78,80,82,85,87–91,93,95,97,100,101,104–107,109–114,116,117,119,120,122–126,129,134–136,138,139,142,146–151,154–156,159–162,167,172,174,180,181,185–187,190–192,196–199,205,206,209,211,213,214,216,217,219,225,229–232,235,237,239,241–243,254,256,260,264,265,268–270,272,275,277,283,286,287,290–292,299,301,302), and knowledge (174 of 242 articles [71.9%]20–22,25–33,36,37,39,42,43,49,53–59,61,62,64–66,68–70,73,76,78,81,82,84,85,91–93,95,97–100,102–107,109,110,113,114,116–121,126,128,130,131,136,138,139,141–143,147,149–152,154,155,157,159–161,163,166–168,170–174,176–180,182–186,188,190–194,197–204,206–210,212–215,219,221,224,226,236–238,241,242,244,246,247,251,252,256–258,264,266–269,271,273,274,276–281,283–288,292–295,297–302); and social influences (121 of 184 articles [65.8%]26,27,30–32,41,42,46,47,49,51,57,58,60,62,63,68,71,77,79,80,82,83,88,90,92,95,99,101,102,106–110,112–114,118,121–124,126,127,129,134–138,146,147,151,153,155,157–159,161,165,167,169,170,176,177,180,182,185,189,195,197–208,210–212,216,217,219,221,223,227,228,233–235,238,242,245,247,249,254,255,257,260,261,264,266,268,269,282,283,286,287,289,291,296–298,301,302) (Table 2). We conducted bivariate analyses of reasons for reluctance and specialty, drug type, intervention, and time (Table 2; eFigure 3 in Supplement 1). Too few studies of recovery support existed to conduct a bivariate analysis with reasons for reluctance. Analysis of combinations of the top 4 reasons for reluctance found the most often paired reluctance reasons were knowledge and skill (135 of 221 articles [61.1%]20–22,25–33,37,39,49,53–55,58,59,61,64–66,68,76,78,81,82,84,85,91–93,95,97–100,102–107,109,110,113,114,116–121,130,131,136,138,139,142,143,147,149,150,152,154,159–161,167,168,172–174,176,178,180,182,183,186,188,190,191,193,194,197–202,204,206–210,213,214,219,221,224,226,236,238,241,242,246,247,256,264,266,268,269,271,273,274,276–279,281,283,285–287,292–295,297,298,301,302), followed by cognitive capacity and institutional environment (99 of 165 articles [60.0%]22,25,26,30,32,37,40,41,47,49,55,58–61,63,64,66,68,74,75,77,78,80,82,87,89–91,93,95,97,100,104,106,107,109,110,112–114,116,117,122–124,126,129,134–136,138,139,146–148,150,151,154,155,159,161,162,167,172,174,180,185,186,192,198,199,206,209,211,216,217,219,229,230,232,239,241–243,260,264,265,268,269,272,275,277,287,290,291,299,301,302) (Table 3). Institutional environment appeared in combination with other reasons more often than any other reason (7 of 12 pairings). Reasons not in our data extraction template were described in a few studies, including lack of demand (13 articles87,92,112,122,143,167,171,214,216,232,257,280,292), cost to the patient (8 articles58,69,148,155,171,174,288,292), and patient refusal (6 articles61,146,170,174,182,206). Analysis of the trend over time for each reason for reluctance revealed a significant increase in identification of social influence (F1,20 = 4.91; P = .04) and relationship (F1,20 = 4.54; P = .046) (eFigure 3 in Supplement 1). We extracted exemplar text from included studies for the top 4 reasons for reluctance (Table 4), discussed in the following section.
Institutional Environment
Reasons for reluctance related to the institutional environment included lack of trained staff66,154,167,182,186,207,242,260 or resources to train staff,59,92,221 acceptance of addiction interventions by staff107,259 or leadership,57,80,155,169,175,261,275 and clinician backup.54,56,64,75,76,90 Regulatory and liability concerns were frequently reported,32,35,50,75,76,87,90,99,107,163,165,167,174,245,259,261 as were record-keeping or confidentiality concerns207,259,275 and staff time required for prior authorizations.92 Often mentioned were also cost to the patient or lack of insurance coverage,148,155,170,171,173,174,182 along with medication unavailability at pharmacies95,144,148,170 and the absence of population-specific patient education materials.260,291 Less frequently cited but noteworthy reasons for reluctance include contractual limitations,291 nonexistent or unimplemented treatment algorithms,99,287 mental health programs not accepting patients with addiction,264 addiction treatment programs rejecting patients deemed insufficiently ready to change or having difficulty matching the level of care needed,229 and difficulty obtaining records from addiction treatment programs.107 Reimbursement can be viewed as a component of institutional environment. In the TDF, reimbursement is 1 part of reinforcement as a reason for reluctance (Box). While reinforcement was 1 of the 2 least often identified reasons for reluctance, data specific to reimbursement was extracted because it is a perennial point of concern in adopting evidence-based interventions for addiction. Physician reimbursement was viewed as insufficient to cover both the staff time necessary to intervene in addiction and the expense of additional staff training.174,207,277 Medicaid reimbursement was specifically highlighted as inadequate.186 In some cases, physicians perceived the reimbursement to be inadequate but were not certain of the reimbursed amount.56
In studies identifying lack of knowledge as a reason for reluctance, knowledge was more deficient for treatment than for screening or diagnosis and for drug use more than for alcohol or tobacco use.20,65,70,93,99,102,117,152,194,221,252,273 Physicians were unfamiliar with the evidence for substance use disorders as biomedical conditions,119,138,199,257 harm reduction strategies,58,154 and screening for risky substance use.59,161 Some physicians lacked awareness of the extent of substance use by their patients.256
Physicians reported lacking skills to conduct interventions effective enough to produce behavior change, including counseling21,38,51,59,117,291 and brief intervention.93,209,229 They also described a lack of skill needed to initiate or manage treatment,92,152,221,273 especially for substance use disorders other than alcohol or tobacco.63,194 In some studies, they equated their lack of skill with lack of experience with observing or delivering a substance use disorder intervention under supervision.22,75,91,238,256 Inabilities to assemble or demonstrate naloxone administration devices58,277 or to deliver appropriate training in its use to patients99 were also noted.
Lack of Cognitive Capacity
Lack of cognitive capacity was not often characterized beyond a general sense of overwhelm with clinical tasks (eg, “just too busy”)64,291 and the need to prioritize patients’ competing needs.58,107,109,268 In some cases, physicians perceived intervening in addiction as too time-consuming, both during the appointment and for monitoring,69,87,90,93,287 or that addiction treatment demand would be too great.66,75,91 Even delegating screening to other clinical team members was viewed as diverting time from the physician visit229; available tools were considered time-consuming.260 Some physicians expected meeting the care needs of patients with addiction to be too time-consuming.
We analyzed 4 main themes related to facilitators. First, physicians need the knowledge and skills to intervene; they need adequate education and training in areas like managing pharmacology. Second, intrapersonal and interpersonal factors exist that facilitate physician intervention. Intrapersonal factors include physician characteristics (eg, work experience, confidence, and practice type) and motivation (eg, desire to improve patient outcomes, reimbursement, and understanding addiction as within their scope of practice). Interpersonal factors include the physician-patient relationship, specifically the patient characteristics that may compel the physician to intervene (eg, the patient is receptive to help). Third, an infrastructure is needed that supports physician interventions and includes institutional changes at the practice level to implement protocols to standardize care (eg, screening and improved technology). An environment that fosters collaboration with other professionals or entities (eg, multidisciplinary teams and referral systems) and offers resources that would support the intervention (eg, materials or tools for use with patients, follow-up care, or treatment facilities) is also essential. Finally, regulation reforms (eg, eliminating prior authorization requirements, expanding substance use disorder insurance coverage, and simplifying laws and policies governing prescribing and medication distribution to patients) would facilitate physician intervention.
The number and growth of publications meeting inclusion criteria for this systematic review demonstrates increasing interest in the perceived and actual barriers to physician engagement with addiction in clinical practice. The significant increase in social influence and relationship as reasons for reluctance over time may indicate increased awareness of stigma and associated social harms. Regarding intervention types, the availability of effective alcohol use disorder and opioid use disorder pharmacotherapies likely accounts for the literature’s focus on those therapies, corresponding with efforts to increase access to medications for opioid use disorder and to promote the adoption of screening, brief intervention, and referral to treatment practices. As the evidence base for a wider array of harm reduction strategies grows, it will be important to understand and address physicians’ perceived and actual barriers to their acceptance and adoption of those strategies. Information is limited on the adoption of recovery support interventions by physicians, a finding that also merits investigation.
That institutional environment is associated with physician reluctance to intervene may not surprise practicing clinicians. The pairing of institutional environment and cognitive capacity may signify the cost in time physicians expend overcoming institutional barriers to EBP for addiction (eg, inefficient workflows and communication and coordination of care across silos). The association of institutional environment with treatment and opioids may reflect the push to increase buprenorphine access despite regulatory impediments and health systems being unprepared for this responsibility.
Strategies to reduce physician reluctance related to institutional environment include greater commitment by health systems to make essential workflow and staffing changes, the breaking down of barriers between addiction services and both medical and mental health care, and commitment by insurers to provide reimbursement that covers the actual cost of providing addiction interventions. The analysis of facilitators supports a specific need for protocols to adequately intervene with patients with either at-risk substance use or substance use disorders. Institutional environment changes (eg, investing in staffing and staff training, implementing standard practices or protocols, and conducting addiction-specific quality assurance) could also facilitate intervention.
Lack of knowledge and skill are top reasons for reluctance, both separately and combined. It is unclear whether survey respondents understood knowledge and skill as the researchers intended because these terms were rarely defined in the studies. Only a few studies allowed for future replication by including objective measures of knowledge or skill (eg, counting successfully delivered services and interviewing patients).
True lack of knowledge and skill can be understood in several ways, including as a manifestation of the volume of information practicing clinicians are required to possess, acquire, and update. For example, physicians need updated information on dosing, pharmacology, and overall efficacy of interventions and medications. This challenge is made harder if interventions (eg, screening practices, initiating pharmacotherapy) are insufficiently adapted for different practice settings. Delivering these interventions effectively, efficiently, and in a nonstigmatizing manner requires skill mastery. Physicians, like other clinicians, acquire their skills by observing and then practicing under supervision. Medical education and postgraduate training have only recently begun to prepare physicians for these tasks.303,304
Ongoing training is critical for physicians to acquire and apply advanced skills in the care of this patient population,305–307 but few opportunities exist to observe and be observed practicing new skills once required medical training is complete. The analysis of facilitators suggests skill training should focus on brief intervention (eg, screening or assessment) and on communication with patients. Trainings accessible to physicians (eg, free or incentivized, hands-on, or delivered in clinical settings) and delivered by specialized trainers and/or mentors would facilitate the growth of a pool of experts to intervene in substance use. Physicians who expand their knowledge and skills should be eligible for continuing medical education credits and increased compensation.
Other reasons for reluctance (eg, negative social influences, negative emotions toward people who use drugs, and fear of harming the relationship with the patient by discussing substance use) could each be viewed as manifestations of stigma associated with substance use disorder and its treatment. Lack of demand may also reflect stigma if it is a manifestation of unwillingness on the part of patients to seek help due to fear of social, legal, and moral judgement or a presumption by the physician that there is no addiction in their community.
These reasons may diminish if effective public and professional education, in particular those developed and led by patient groups or by people who use drugs,308–312 are delivered to counter stigma.313 The analysis of facilitators suggests the following may be helpful: educational materials for patients and families, community outreach, and public health campaigns promoting nonstigmatizing language.
Reducing stigma will not be enough to address fear of harming the patient relationship, especially for physicians who care for minors and other populations that may be subject to punitive consequences of addiction. These physicians must consider additional confidentiality requirements, and their fear of harming the patient by triggering negative social and legal consequences may be more of a deterrent than previously considered. Interpersonal aspects of the patient-physician relationship and how they create reluctance or facilitate intervention are not well understood, although the analysis of facilitators shows that physicians may be motivated to intervene in substance use disorders when they have an established relationship with the patient, the patient is receptive to help, and/or the desire to improve patient outcomes is strong. Future research should examine unintended impacts of increased physician intervention in addiction like strain on the physician-patient relationship, less opportunity to meet other health care needs, and stigmatizing interactions with other health care clinicians due to the substance use disorder diagnosis being more widely documented.
This study has limitations. Inconsistent use of terms across included studies increased the complexity and interpretation of this analysis, but analysis of a sample this size can still inform research and policy. Studies were often developed without the benefit of a theoretical framework. Survey development lacked or failed to report participation of the audience of focus and/or was not piloted, raising concerns about the validity and applicability of results. During the years this systematic review covered, new medications and formulations became available, making comparison across decades challenging. The unregulated drug market also evolved, resulting in changes to illicit substances, methods of using them, and the regulatory environment in which clinicians address substance use. This review was limited to physicians, some of whom may have participated in more than 1 survey or focus group in the included studies. Although the results are relevant to the practice environment of many clinicians, including those specializing in addiction, they do not reflect the unique challenges that may be encountered by specific disciplines. Although we collected and described data about facilitators, the original search was not designed specifically to retrieve publications about facilitators of intervention in addiction.
These data suggest that policy, regulatory, or accreditation changes are needed to systematically address institutional barriers, as well as increases to physician reimbursement and opportunities for clinically relevant training that provides both skill development and knowledge gain. Another systematic review of facilitators and reluctance among other clinical disciplines may refine the recommendations presented here. Future studies of clinician reluctance to adopt EBPs for addiction need to be of higher quality. They, at a minimum, should employ a theoretical framework and adhere to survey development best practices or use a validated survey instrument.
Accepted for Publication: May 7, 2024.
Published: July 17, 2024. doi:10.1001/jamanetworkopen.2024.20837
Source: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2821497