Guideline Statements and Implementation

Implementation

For any patient who is undergoing an initial psychiatric evaluation, it is important to assess the patient’s use of tobacco, alcohol, and other substances, as well as any misuse of prescribed or over-the-counter (OTC) medications or supplements (see Guideline II, “Substance Use Assessment,” in the APA Practice Guidelines for the Psychiatric Evaluation of Adults; American Psychiatric Association 2016). In individuals with AUD, both the 12-month and lifetime odds ratios (ORs) of nicotine use and other substance use disorders are increased (Grant et al. 2015), which supports the need to inquire about past as well as current use. In addition, knowledge of past and current use can inform treatment planning. Information can be obtained through face-to-face interviews, standardized assessment tools, laboratory testing, and input from collateral sources such as family members, other health professionals, medical records, history of electronic prescriptions, or prescription drug monitoring program data.

In face-to-face interviews with the patient, a nonjudgmental and open-ended approach to questions is typically most informative. The interviewer can begin by seeking the patient’s permission to ask about or discuss alcohol and other substance use before actually doing so, respecting and documenting the wishes of patients who do not choose to discuss this information, and speaking openly with the patient about the confidentiality of the information and any limits on confidentiality that may exist. Questioning and terminology should be adapted to the individual patient on the basis of such factors as age and culture. The specific substances that are asked about will vary with the clinical context and may include, but are not limited to, alcohol; caffeine; cannabis; hallucinogens; inhalants; opioids; sedatives, hypnotics, and anxiolytics; stimulants, including amphetamine-type substances, cocaine, and other stimulants; tobacco; and other substances. Questions about misuse of prescribed or OTC medications or supplements can often be introduced while the clinician is taking a history of the patient’s prescribed medications. Depending on the substance(s) being used, additional follow-up questions will generally be needed to delineate the route, quantity, frequency, pattern, typical setting, and circumstances of use as well as self-perceived benefits and psychiatric and other consequences of use. In terms of alcohol use, it can be helpful to identify the type of alcohol used (e.g., beer, wine, distilled spirits).

For a variety of reasons (e.g., stigma, memory impairment, potential for negative consequences), individuals may underreport the type or extent of alcohol or other substance use. In addition to information about use that can be gained from collateral sources or laboratory studies, aspects of the patient’s history may signal a need to probe further in identifying problematic alcohol or other substance use. For example, additional questioning may be needed to explore issues such as family discord; academic or occupational problems; difficulties with mood, sleep, or sexual functioning; or specific physical symptoms (e.g., gastrointestinal distress) or symptom patterns (e.g., anxiety or headaches after every weekend). Observations made during the interview can provide additional clues to possible use (e.g., an odor of cigarettes or alcohol on the patient’s breath, physical signs of injection drug use, slurred speech, tremulousness or other evidence of alcohol or substance intoxication or withdrawal).

Information from self-report rating scales can complement information from the face-to-face interview (Guideline II, American Psychiatric Association 2016). The DSM-5 Self-Rated Level 1 Cross-Cutting Symptom Measure (available online at http://www.psychiatry.org/practice/dsm/dsm5/online-assessment-measures) permits initial screening; patients can be asked for additional details on substance use items through administration of the DSM-5 Level 2—Substance Use measure (American Psychiatric Association 2013).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

As described in the APA Practice Guidelines for the Psychiatric Evaluation of Adults (American Psychiatric Association 2016), individuals who were identified by peers as experts in psychiatric evaluation assessed patients for use of alcohol or other substances at consistently high rates, whereas assessment of past and current tobacco use were also high but showed opportunity for improvement. The typical practices of other psychiatrists and mental health professionals are unknown, but rates of tobacco use screening have been declining among psychiatrists practicing in ambulatory settings (Rogers and Sherman 2014). Data from ambulatory settings (Glass et al. 2016) suggest that many individuals receive screening for alcohol use, but approximately one-third of individuals do not. Rates of screening for use of other substances, including misuse of prescribed or OTC medications, are likely to be less than rates of screening for either tobacco or alcohol use.

Several existing measures are of relevance to this recommendation. National Quality Forum (NQF) Measure 110, “Bipolar Disorder and Major Depression: Appraisal for Alcohol or Chemical Substance Use,” assesses the percentage of patients with depression or bipolar disorder, with evidence of an initial assessment that includes an appraisal for alcohol or substance use (www.qualityforum.org/QPS/0110). In terms of tobacco use, the NQF-endorsed Measure 0028, “Preventive Care & Screening: Tobacco Use: Screening & Cessation Intervention,” assesses the percentage of adult patients who are screened every 2 years for tobacco use and who receive cessation counseling intervention if identified as a tobacco user (www.qualityforum.org/QPS/0028). Several other NQF-endorsed treatment performance measures are related to screening for tobacco use in inpatient settings. In addition, an expert panel convened by the RAND Corporation has suggested that detailed specification development and pilot testing would be appropriate for a potential AUD-related quality measure on screening for substance use, including tobacco use, in individuals with an Alcohol Use Disorders Identification Test–Concise (AUDIT-C) score of greater than or equal to 5 (Hepner et al. 2017). The American Society of Addiction Medicine also has proposed a measure on screening for tobacco use disorder (American Society of Addiction Medicine 2014). Before adopting any measures, it is important to determine whether the measure has been validated in the population and setting of interest. Thus, it is recommended at this time that only measures specified or endorsed for outpatients be used in that treatment setting.

The most effective manner to assess and report on measures related to substance use is unclear. Several options for reporting are in practice, and have been proposed.

As described in the APA Practice Guidelines for the Psychiatric Evaluation of Adults (American Psychiatric Association 2016), a comprehensive measure could be derived that assesses the percentage of patients seen in an initial evaluation who are screened for the use of tobacco, alcohol, or other substances as well as for the misuse of prescribed or OTC medications.

Because existing measures already include a tobacco use screening measure, it may be preferable to focus new measure development on assessment of current and past alcohol use. Such a measure could be paired with a distinct measure on assessment of substance use. Alternatively, a measure on the assessment of alcohol use could be paired with a measure that determines whether treatment for AUD was initiated.

In practices that use an electronic health record, a measure on the assessment of past and current alcohol use could be implemented by measuring for the presence or absence of text in corresponding fields labeled “past alcohol use” and “current alcohol use.” This approach would aim to ensure that assessment has occurred and is documented in a patient’s record but would allow for maximum flexibility in how clinicians document findings of their assessments without endorsing use of a specific scale or method of assessment. Regardless of the approach that is chosen, quality improvement activities derived from this recommendation, including performance measures, should not oversimplify the process of assessing alcohol use, as alcohol use is commonly underreported by patients and often requires use of clinical interviewing skills to elicit accurate information. Exceptions to the denominator of the measure should be specified and might include individuals who are unable to participate in the evaluation because of their current mental status. Other exceptions might also be appropriate.

Statement 2: Use of Quantitative Behavioral Measures

Implementation

Quantitative behavioral measures should be used during the initial psychiatric evaluation of a patient with AUD to detect the presence of alcohol misuse and determine its severity. The intent of using a quantitative behavioral measure is not to establish diagnosis but rather to complement other aspects of the screening and assessment process. Depending on the measure, it can also serve as a baseline measure to judge the effects of treatment. Co-occurring psychiatric conditions or cognitive impairment may limit some patients’ ability to complete self-report instruments. In these circumstances, it may be necessary to place greater reliance on collateral sources of information such as family members or staff members of sober living homes or community residence programs, if applicable.

A number of validated scales and screening tools have been developed (Jonas et al. 2012a, 2012b). Although recommending a particular scale is outside the scope of this practice guideline, considerations in choosing a scale include the age of the patient, clinical setting, time available for administration, and therapeutic objective (i.e., screening vs. diagnosis vs. ongoing monitoring). For example, the CAGE questionnaire (Ewing 1984) has been studied as a screening tool for AUD and asks whether the individual felt a need to cut down on drinking, was annoyed by criticism of drinking, felt guilty about drinking, or had a morning eye-opener. CAGE does not provide enough information to suggest a diagnosis of AUD or to be used in monitoring alcohol use in patients with known AUD (do Amaral and Malbergier 2008). In addition, it is less sensitive in screening for mild to moderate AUD than are some other measures. The CRAFFT screening tool is intended to be developmentally appropriate for adolescents and includes questions about being in a car driven by someone who was using alcohol or drugs, use of alcohol or drugs to relax or when alone, forgetting what was done while using alcohol or drugs, being told by family or friends to cut down on use, and getting into trouble while using alcohol or drugs (Knight et al. 1999). On the other hand, the AUDIT (Saunders et al. 1993) and its shortened form, the AUDIT-C (Bush et al. 1998), are more appropriate for use with adult patients, including pregnant women (Burns et al. 2010). With both the AUDIT and the AUDIT-C, interpretation of the resulting score differs by sex and by age, with women and individuals age 65 and older having lower score thresholds than men or adults under age 65. The National Institute on Alcohol Abuse and Alcoholism (NIAAA) one-question screen, which is useful in identifying problematic drinking in primary care settings (Smith et al. 2009), asks, “How many times in the past year have you had X or more drinks in a day?”, where X is 5 for men and 4 for women, with a response greater than 1 constituting a positive screen (U.S. Department of Health and Human Services 2007). Many other measures are also available that may be useful for specific practice settings or individual patients (see Deady 2009 for review).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

It is not known how frequently psychiatrists and other health professionals use a quantitative behavioral measure to detect the presence of alcohol misuse and assess its severity in ambulatory settings. Anecdotal observations suggest variability in the routine use of such measures, and even when such measures are used routinely, there can be variability in results (Bradley et al. 2011).

Use of quantitative behavioral measures to assess individuals with AUD could be one approach to meeting a measure on assessing past and current use of alcohol. As described in Statement 1, a measure could consider the presence or absence of scoring from a relevant measurement tool but should avoid endorsing the use of a specific scale.

One example measure is the NQF-endorsed Measure 2152, “Preventive Care and Screening: Unhealthy Alcohol Use: Screening & Brief Counseling.” The measure specifies the use of the AUDIT, the AUDIT-C, or the NIAAA one-question screen. Brief counseling is described as at least one session of “a minimum of 5–15 minutes, which may include: feedback on alcohol use and harms; identification of high risk situations for drinking and coping strategies; increased motivation and the development of a personal plan to reduce drinking” (National Quality Measures Clearinghouse 2016). An expert panel convened by the RAND Corporation has suggested a number of potential AUD-related quality measures that would be appropriate for detailed specification development and pilot testing; many of these use specific threshold scores on the AUDIT-C as a method for identifying individuals who are appropriate for the use of the measure (Hepner et al. 2017).

A process-focused internal or health system–based quality improvement measure could also determine rates of quantitative behavioral measure use and implement quality improvement initiatives to increase the frequency with which such measures are used in individuals with AUD.

Statement 3: Use of Physiological Biomarkers

Implementation

Alcohol consumption can also be evaluated and monitored using alcohol biomarkers (see reviews by the Substance Abuse and Mental Health Services Administration 2012, Dasgupta 2015, and Litten and colleagues 2010).

Biomarkers for alcohol consumption are not intended to replace the clinical interview and quantitative behavioral measures but may augment these assessments (do Amaral and Malbergier 2008; Miller et al. 2004) along with input from collateral informants. Alcohol consumption biomarkers may be particularly useful in certain patient populations, such as those with co-occurring psychiatric illness or cognitive impairment that limits the ability to self-report alcohol use. Biomarker testing may also be of particular use when a clinician suspects a patient to be minimizing reported use of alcohol (e.g., due to concerns about employment or insurance termination), when heavy drinking requires verification (e.g., forensic liability and custody cases), or when abstinence is needed (e.g., in court-mandated alcohol treatment). In addition, some biomarkers can help to evaluate for alcohol-related organ damage, which may prompt treatment referral for medical complications of alcohol use. When biomarkers are used, results should be discussed with patients in ways that encourage open and honest communication about alcohol consumption (Miller et al. 2004).

Biomarkers may be obtained from a variety of sources (e.g., blood, urine, hair). Direct biomarkers measure alcohol or alcohol metabolites over a time course of hours (blood ethanol level) to months (hair ethyl glucuronide [EtG]) and generally are more sensitive to any alcohol consumption. Other direct biomarkers, such as phosphatidylethanol (PEth), detect steady low to heavy drinking over a period of weeks. Carbohydrate-deficient transferrin (CDT), an indirect marker, detects only heavy drinking (e.g., four or more drinks per day for women and five or more per day for men consumed frequently in the weeks prior to testing). In contrast, other indirect biomarkers, such as gamma-glutamyl transferase (GGT), alanine transaminase (ALT), aspartate transaminase (AST), and mean corpuscular volume (MCV), typically reflect organ damage or physiological dysfunction resulting from more chronic, heavy alcohol consumption. Using a combination of biomarkers should improve sensitivity and increase specificity (e.g., a less specific positive GGT would be confirmed as alcohol related by a positive %CDT).

There are several other factors to consider when choosing a biomarker. It is important to evaluate for co-occurring medical conditions or medications that may interfere with biomarker testing (see reviews by the Substance Abuse and Mental Health Services Administration 2012, Dasgupta 2015, and Litten and colleagues 2010). Interpreting biomarker levels is further complicated by variations in assay techniques and threshold values for a positive test (Weykamp et al. 2013). Different thresholds may also be necessary depending on the patient’s therapeutic goal (e.g., abstinence vs. moderation) (Balldin et al. 2010). Access to urine (with risk for obfuscation), blood (e.g., availability of phlebotomy services), and insurance coverage for specific biomarkers can also influence test selection.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

As a suggestion, this statement is inappropriate for use as a quality measure.

Statement 4: Assessment of Co-occurring Conditions

Implementation

AUD frequently co-occurs with other psychiatric disorders, particularly mood or anxiety disorders (Hasin et al. 2005). Identifying co-occurring conditions can aid treatment planning and help in providing integrated care for AUD and other psychiatric conditions. The relationship between alcohol use and psychiatric symptoms is complex and likely bidirectional (Grant et al. 2004; Kenneson et al. 2013; Martins and Gorelick 2011). Alcohol may exacerbate some symptoms (e.g., depressed mood) during periods of use or withdrawal but may reduce the patient’s experience of other symptoms (e.g., anxiety, psychosis), contributing to ongoing alcohol use. Problematic alcohol use may also occur in the context of certain disorders that result in impaired impulse control (e.g., bipolar disorder, borderline personality disorder) or may itself lead to worsening behavioral disinhibition. Therefore, it is important to screen for other co-occurring psychiatric disorders. It is also important to assess a patient’s risk for suicide and aggressive behaviors (American Psychiatric Association 2016; Buchanan et al. 2011) because heavy alcohol use is a known risk factor for both suicide (Norström and Rossow 2016) and violence (Abramsky et al. 2011; Branas et al. 2016). Such assessments can be accomplished through clinical interview, mental status examination, or use of quantitative measures. Additionally, as described above, screening for other substance use disorders is important for treatment planning because co-occurring disorders may influence medication considerations. For example, an individual with comorbid AUD and opioid use disorder might benefit from extended-release naltrexone to treat both disorders after an informed consent discussion that includes the risk of precipitated opioid withdrawal. More detailed recommendations about screening for co-occurring conditions can be found in the APA Practice Guidelines for the Psychiatric Evaluation of Adults (American Psychiatric Association 2016).

It is also important to screen for nonpsychiatric medical conditions that may have arisen as sequelae of or independent from heavy alcohol use. Such assessments include, but are not limited to, measuring serum creatinine and hepatic transaminase levels. One should also evaluate for other causes of hepatic (e.g., viral hepatitis) or renal (e.g., diabetes mellitus, hypertension, HIV) impairment because this may influence the choice of AUD pharmacotherapy. For example, acamprosate is contraindicated in severe renal disease (creatinine clearance [CrCl] < 30 mL/min or estimated glomerular filtration rate (eGFR) < 30 mL/min/1.73 m²), and naltrexone must be used cautiously in individuals with hepatic impairment (see Statement 17: Implementation).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

As described in the APA Practice Guidelines for the Psychiatric Evaluation of Adults (American Psychiatric Association 2016), individuals who were identified by peers as experts in psychiatric evaluation reported high rates of inquiring about co-occurring conditions. The typical practices of other psychiatrists and mental health professionals are unknown. There are many challenges in developing a quality measure from assessment-related recommendations (American Psychiatric Association 2016). There are no NQF-endorsed recommendations on this topic. However, some unendorsed measures exist related to co-occurring conditions in individuals with psychiatric illness. These would be useful to review before considering development of a new measure. In addition, an expert panel convened by the RAND Corporation has suggested screening for liver disease in individuals with an AUD diagnosis or an AUDIT-C score of greater than or equal to 8 (Hepner et al. 2017), and the American Society of Addiction Medicine has proposed a measure on documenting diagnoses of co-occurring psychiatric disorders (American Society of Addiction Medicine 2014). With the increasing use of electronic medical record systems and associated recording of problems and diagnoses using structured terminology, it may be possible to develop electronic measures from this recommendation that could be used for process-focused internal or health system–based quality improvement initiatives.

Statement 5: Determination of Initial Treatment Goals

Implementation

Clinicians should collaborate with patients to identify specific treatment goals regarding their alcohol use. With the patient’s permission, involvement of family members in developing treatment goals can be helpful. Options for treatment goals might include abstinence, reduction in alcohol use, or eliminating drinking in particularly high-risk situations (e.g., at work, before driving, when responsible for caring for children). Data have shown that having explicit drinking goals at baseline may be associated with improved AUD treatment outcomes (Dunn and Strain 2013). Abstinence as a pretreatment goal has been associated with greater rates of abstinence or moderation, but all groups with an explicit pretreatment goal showed some reduction in alcohol use. Abstinent and nonabstinent drinking goals can include controlled or occasional use, abstinence with the recognition that slips may occur, or total abstinence on a short- or long-term basis (Dunn and Strain 2013).

Motivational interviewing (MI) is one model for having discussions about goals with patients (Levounis et al. 2017; Miller and Rollnick 2013). In MI, the clinician first asks permission to discuss alcohol use. After the patient consents, the goal is to help the patient articulate his or her ambivalence about drinking by asking about positive and negative aspects of alcohol use along with assessments of readiness to reduce drinking and confidence in his or her ability to do so. Such discussions are facilitated by a clinician stance that is curious and nonjudgmental, while also expressing concern for the patient’s well-being.

Clinicians should clearly document the agreed-on treatment goals in the medical record (e.g., a brief notation as part of a progress note). Additional documentation may be needed when the goals of the patient and the clinician are not in agreement. For example, a patient may only agree to a reduction in drinking but continue to drink in situations that place him or her at risk of legal involvement (e.g., DUIs, DWIs) or of significant medical sequelae from alcohol use (e.g., hepatic injury). Progress note documentation should reflect that both the clinician and patient understand these risks and have engaged in a discussion about them. As the evaluation and treatment of the patient proceed, the patient and clinician can adjust these initial goals on the basis of factors such as responses to treatment, additional history, family input, or education about treatment options and potential treatment effects (e.g., reduced craving).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

As a suggestion, this statement is inappropriate for use as a quality measure. A process-focused internal or health system–based quality improvement measure could determine rates at which initial treatment goals are documented. Quality improvement initiatives could be implemented to increase the frequency at which such discussions and documentation occur in individuals with AUD.

Statement 6: Discussion of Legal Obligations

Implementation

Some patients come to treatment as a consequence of legal involvement, and their engagement in treatment may be court mandated. The initial assessment of AUD should include inquiry about legal involvement and legal obligations, if any, that the patient may have in relation to alcohol use. For individuals in mandated treatment, reporting requirements will vary with the local jurisdiction but should be discussed with the patient. Mandated treatment situations may also influence the treatment goals (e.g., abstinence) and the monitoring of abstinence, such as with serum ethanol levels, ethanol breath tests, or other alcohol-related biomarkers. It is important to document any such legal obligations in the medical record (e.g., as a brief notation as part of a progress note) along with a discussion of the treatment plan and therapeutic goals.

Balancing of Potential Benefits and Harms in Rating he Strength of the Guideline Statement

Quality Measurement Considerations

As a suggestion, this statement is inappropriate for use as a quality measure. A process-focused internal or health system–based quality improvement measure could determine rates at which initial treatment goals are documented, including discussion of legal obligations, if any. Quality improvement initiatives could then be implemented to increase the frequency at which such discussions and documentation occur in individuals with AUD.

Statement 7: Review of Risks to Self and Others

Implementation

Discussion of risks to self and others from continued alcohol use will be a natural outgrowth of the assessment. Most individuals who are seeking treatment will already have experienced some negative consequences of alcohol use, which they will typically mention in the context of describing current motivations for treatment. Additional risks can be explored with the patient and documented (e.g., a brief notation as part of a progress note), with the aim of reducing harms associated with drinking. Health risks can include increases in all-cause mortality (Laramée et al. 2015); injury (Cherpitel et al. 2017) or physical or psychological problems (Borges et al. 2017; Rehm et al. 2013, 2017; Shield et al. 2013) related to alcohol use; or interactions between alcohol and other medications that the patient is taking (Breslow et al. 2015). Other common risks include difficulties in occupational, academic, family, social, or other interpersonal functioning; legal involvement; or use of alcohol in physically hazardous situations. When discussing potential harms of alcohol use, it is also important to consider that risk may vary with factors such as concomitant health conditions, age (Moore et al. 2006), and sex and gender (Erol and Karpyak 2015). Materials available through NIAAA may also be helpful in discussing risks of AUD with patients and families (U.S. Department of Health and Human Services 2007). Screening instruments, such as the Drinker Inventory of Consequences (Miller et al. 1995) or the shortened version, the Short Index of Problems (SIP; Feinn et al. 2003; Forcehimes et al. 2007), may also aid clinicians in identifying and supporting discussions of negative consequences of alcohol use.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

As a suggestion, this statement is inappropriate for use as a quality measure. A process-focused internal or health system–based quality improvement measure could determine rates at which the risks of alcohol use have been discussed and documented. Quality improvement initiatives could be implemented to increase the frequency at which such discussions and documentation occur in individuals with AUD.

Statement 8: Evidence-Based Treatment Planning

Implementation

In treating individuals with AUD, a person-centered treatment plan should be developed, documented in the medical record (e.g., as part of a progress note), and updated at appropriate intervals. Such a plan may require tailoring based on sociocultural factors such as gender and age (Erol and Karpyak 2015; Kerr-Corrêa et al. 2007; Sudhinaraset et al. 2016). A person-centered treatment plan can be recorded as part of an evaluation note or progress note and does not need to adhere to a defined development process (e.g., face-to-face multidisciplinary team meeting) or format (e.g., time-specified goals and objectives). However, it should give an overview of the identified clinical and psychosocial issues along with a specific plan for addressing factors such as acute intoxication or alcohol-related medical issues (if present), further history and mental status examination, physical examination (either by the evaluating clinician or another health professional), laboratory testing (as needed, based on the history, examination, and planned treatments), ongoing monitoring, and nonpharmacological and pharmacological interventions, as indicated (Substance Abuse and Mental Health Services Administration and National Institute on Alcohol Abuse and Alcoholism 2015). Plans can also include educating patients about treatment options, engaging family members, collaborating with other treating clinicians, or providing integrated care. Depending on the urgency of the initial clinical presentation, the availability of laboratory results, or collateral informants, the initial treatment plan may need to be augmented over several visits and as more details of history and treatment response are obtained. Collateral informants such as family members, friends, or other treating health professionals may express specific concerns about the individual’s alcohol use or related behaviors or concerns or biases about specific treatment approaches. If present, such concerns should be documented and addressed as part of the treatment plan. Additionally, the patient’s goals and readiness to change his or her alcohol consumption may evolve over time and necessitate changes to the treatment plan. Changes to the treatment plan will also be needed if a patient has not tolerated or responded to a specific treatment or if he or she chooses to switch treatment approaches. For example, if a patient does not wish to receive further nonpharmacological treatment, a reconsideration of AUD pharmacotherapy would be warranted if such medications are not already prescribed.

As part of a person-centered treatment plan, it is important to consider both nonpharmacological and pharmacological treatment approaches. Although recommending a particular nonpharmacological approach is outside the scope of this practice guideline, there are several evidence-based options for the treatment of AUD. These include MET (Lenz et al. 2016) and CBT for AUD (Epstein and McCrady 2009). MET is a manualized psychotherapy based on the principles of motivational interviewing that has been shown in multiple studies to have a small to medium effect size on achieving abstinence (Dieperink et al. 2014; Lenz et al. 2016). This treatment is designed to help patients develop intrinsic motivation to reduce or abstain from alcohol use by helping them explore their own ambivalence about alcohol use and its sequelae. Motivational interviewing principles can also be used to foster shared-decision making regarding AUD pharmacotherapy and in promoting medication adherence (Levounis et al. 2017). CBT focuses on the relationships between thoughts, feelings, and behaviors (Epstein and McCrady 2009). Particular attention is paid to strategies that help the patient manage urges and triggers (i.e., cues) to drink. Medical management (MM) is also a manualized treatment (Pettinati et al. 2004) that was developed for use in the Combined Pharmacotherapies and Behavioral Interventions for Alcohol Dependence (COMBINE) study. It provides education and strategies to support abstinence and promote medication adherence. Community-based peer support groups such as Alcoholics Anonymous (AA) and other 12-step programs have been helpful for many patients. In fact, TSF relies on the utility of community-based peer supports (Kaskutas 2009; Kaskutas et al. 2009; Kowinski et al. 1992; Project MATCH Research Group 1998a, 1998b). Nevertheless, the focus and structure of groups can vary considerably, and there is a paucity of research on these modalities (Ferri et al. 2006). For these reasons, community-based peer support programs can assist many individuals in achieving long-term remission from AUD but cannot substitute for formal medical treatment in the management of AUD.

When an individual with AUD also has other psychiatric conditions (including other substance use disorders), the treatment plan should assure that each of the co-occurring disorders is addressed, either individually (with appropriate care coordination) or using an integrated model of care. The statements in this guideline should generally be applicable to individuals with co-occurring conditions, although the evidence base in individuals with co-occurring disorders remains limited (Tiet and Mausbach 2007; see Appendix B).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

It is not known whether psychiatrists and other mental health professionals typically develop and document a comprehensive and person-centered treatment plan that includes evidence-based nonpharmacological and pharmacological treatments, and there is likely to be variability. Among individuals who were identified with AUD via screening in general ambulatory settings, only a small fraction received any information about treatment (Glass et al. 2016). Some specific elements of a comprehensive treatment plan, including counseling about treatment options, offer of psychotherapy, receipt of AUD pharmacotherapy, referral to community-based recovery support, and integrated treatment of co-occurring disorders, have been suggested for detailed development of quality measure specifications and pilot testing by an expert panel convened by the RAND Corporation on AUD-related quality measures (Hepner et al. 2017). Nevertheless, an overarching performance measure derived from this recommendation is not recommended because of the associated burdens and practical challenges. Clinical judgment would be needed to determine whether a documented treatment plan was comprehensive and person centered, even if listed treatments were evidence based. If a performance measure assessed for the presence or absence of specific text in the medical record, increased documentation burden could result. Such an approach could also foster overuse of standardized language that would not accurately reflect what has occurred in practice.

Implementation

Naltrexone and acamprosate have the best available evidence as pharmacotherapy for patients with AUD (Center for Substance Abuse and Treatment 2009; Jonas et al. 2014). In most studies, participants were included on the basis of a DSM-IV diagnosis of alcohol dependence, which roughly corresponds to moderate to severe AUD in DSM-5 (Compton et al. 2013; Hasin et al. 2013; Peer et al. 2013). Use of these medications may also be appropriate to consider on an individualized basis for patients with mild AUD, particularly if the patient prefers this treatment modality.

Acamprosate is efficacious in the treatment of AUD when administered at a mean dose of 1998 mg per day, typically 666 mg three times per day (Jonas et al. 2014). Although its exact mechanism of action is unclear, it may act by modulating glutamate, with indirect effects on other neurotransmitters or ion channels (Kalk and Lingford-Hughes 2014). Individuals who were randomly assigned to take acamprosate were significantly less likely to return to drinking after attaining abstinence and had a significant reduction in the number of drinking days, although data on the number of heavy drinking days were mixed. Most experts recommend starting treatment as soon as abstinence is attained and continuing even if the patient relapses (Jonas et al. 2014). The most positive data for acamprosate efficacy comes from outside the United States, where it is typically started in the hospital after detoxification and a period of abstinence.

The lack of metabolism of acamprosate through the liver and the lack of reported hepatotoxicity with acamprosate (National Library of Medicine 2017a) are often important considerations in its use given the significant rates of hepatic dysfunction in individuals with AUD (O’Shea et al. 2010). However, because of the excretion of acamprosate through the kidneys, serum creatinine should be measured at baseline and, in individuals with a history of renal impairment, results should be reviewed before initiating treatment. Acamprosate is contraindicated if estimated CrCl is less than 30 mL/min or eGFR is less than 30 mL/min/1.73 m²; dose reduction may be necessary for CrCl values between 30 and 50 mL/min or eGFR values between 30 and 59 mL/min/1.73 m². Common side effects include diarrhea (17% compared with 10% in placebo; Micromedex 2017a).

Naltrexone is an opioid receptor antagonist that has efficacy in the treatment of both AUD and opioid use disorder. It has greatest affinity for μ opioid receptors, next highest affinity for δ opioid receptors, and lowest affinity for κ opioid receptors (Ashenhurst et al. 2012). This medication has been associated with a reduced likelihood of return to drinking and with fewer drinking days overall. Naltrexone is also thought to decrease the subjective experience of “craving” (Maisel et al. 2013). Naltrexone is available in both a daily oral and monthly depot intramuscular (im) injection. Although long-acting injectable naltrexone may improve adherence (Hartung et al. 2014), there have been no head-to-head comparisons of oral versus injectable naltrexone for AUD, and both formulations appear to be effective. Engaging family members or others to assist with adherence can be particularly helpful if the oral formulation of naltrexone is used. The recommended dose of oral naltrexone is 50 mg daily; however, some patients may require doses up to 100 mg daily to achieve efficacy, which was the dose of naltrexone used in the COMBINE trial (Anton et al. 2006; Garbutt et al. 2005; McCaul et al. 2000a, 2000b). For long-acting naltrexone, the dose is 380 mg im every 4 weeks. There is limited information on use of naltrexone in individuals with co-occurring disorders, but some studies suggest benefit for combined treatment with naltrexone and an antidepressant in depression (Pettinati et al. 2010) and in posttraumatic stress disorder (PTSD; Petrakis et al. 2012) that co-occurs with AUD. Also, smokers with AUD may respond better than nonsmokers to naltrexone (Fucito et al. 2012), which could also influence treatment selection.

Naltrexone is generally well tolerated in clinical trials. Potential gastrointestinal side effects of naltrexone may occur more often among women than men (Herbeck et al. 2016) and include abdominal pain (11% vs. 8% in placebo), diarrhea (13% vs. 10% in placebo), nausea (29% vs. 11% in placebo), and vomiting (12% vs. 6% in placebo; Micromedex 2017c). Dizziness also appears to be more frequent with naltrexone (13% vs. 4% in placebo; Micromedex 2017c). In clinical trials of oral and long-acting injectable naltrexone, rates of anxiety and depression were comparable for naltrexone-treated individuals as compared with placebo. Suicide, suicide attempts, and suicidal ideation were reported in postmarketing surveillance but were infrequent in clinical trials (1% with long-acting injectable naltrexone vs. 0% with placebo; 0%–1% with oral naltrexone vs. 0%–3% with placebo in an open-label trial; Micromedex 2017c). For individuals treated with long-acting injectable naltrexone, pain or induration can occur at the injection site. The potential for bleeding at the injection site should be taken into consideration for patients who have coagulopathy or are taking anticoagulants.

Hepatic functioning can also be affected by naltrexone, and the labeling includes a warning about use of this medication in patients with acute hepatitis or liver failure. With naltrexone, assessment of liver chemistries is appropriate prior to treatment with additional evaluation or consultation and follow-up liver chemistries obtained, as indicated, depending on the extent of any abnormalities (Kwo et al. 2017). The American College of Gastroenterology Clinical Guideline Evaluation of Abnormal Liver Chemistries suggests additional history, physical examination, and laboratory assessment for elevations of AST and ALT that are borderline (less than twice the upper limit of normal) or mild (two to five times the upper limit of normal), with assessment for signs of acute liver failure at values of AST and ALT that are more than five times the upper limit of normal (Kwo et al. 2017). In clinical trials, individuals were generally excluded if hepatic enzyme levels were more than three times the upper limit of normal. In the COMBINE study (Anton et al. 2006), there was an increase in AST or ALT to more than five times the upper limit in 0 of 309 placebo subjects (0%) and 1 of 303 (0%) acamprosate subjects as compared with 6 of 309 naltrexone-treated subjects (2%) and 5 of 305 subjects (2%) treated with acamprosate and naltrexone (p = 0.02). However, other studies have suggested comparable rates of elevations in hepatic enzymes with naltrexone as with placebo, even in patient populations at increased risk for hepatic dysfunction due to co-occurring hepatitis C or HIV infection (Croop et al. 1997; Lucey et al. 2008; M. C. Mitchell et al. 2012; Tetrault et al. 2012; Vagenas et al. 2014).

Because naltrexone is an opioid receptor antagonist, naltrexone may lead to reduced effectiveness of opioids taken for analgesia. Additionally, depending on the half-life of the opioid consumed, outpatients must be abstinent from opioids for 7–14 days prior to starting naltrexone and should be informed of the risk for precipitating opioid withdrawal if naltrexone is used in conjunction with an opioid. If prescription-related information is available through an electronic medical record or prescription drug monitoring program, it should be checked for current or recent opioid prescriptions. Coordinating care with other clinicians is also important. Some clinicians suggest obtaining urine toxicology screening to confirm the absence of opioids before starting naltrexone, particularly if use of the long-acting injectable formulation is planned. It is also advisable for patients to carry a wallet card noting that they are taking naltrexone so this information will be available to emergency personnel. A template for wallet cards and sample templates for documenting medication management visits are available through NIAAA (U.S. Department of Health and Human Services 2007).

In choosing between acamprosate and naltrexone for an individual patient, selection of a medication is likely to be guided by factors such as ease of administration, available formulations, side effect profile, potential risks in women who are pregnant or breastfeeding (see guideline Statement 14), the presence of co-occurring conditions (e.g., hepatic or renal disease), or the presence of specific features of AUD (e.g., craving). In the COMBINE study, patients who received naltrexone and medical management had a greater proportion of days abstinent and a reduced risk of having a heavy drinking day, whereas acamprosate did not affect drinking outcomes in any of the treatment arms (Anton et al. 2006). However, the German PREDICT study (Mann et al. 2013) found no difference among naltrexone, acamprosate, and placebo groups on the time to first heavy drinking. Consistent with this, the AHRQ systematic review and meta-analysis (Jonas et al. 2014) found no statistically significant difference between naltrexone and acamprosate in the percent with a return to any drinking, the percent with a return to heavy drinking, or the number of drinking days, suggesting that neither of these medications was superior to the other. Thus, either naltrexone or acamprosate could be viewed as an appropriate initial treatment choice, depending on other patient-specific considerations. Decisions about the duration of treatment with these medications will also be based on individual factors such as patient preference, disorder severity, history of relapses, potential consequences of relapse, clinical response, and tolerability. There is insufficient evidence available on concomitant use of acamprosate and naltrexone to determine the benefits and harms of combined treatment or to make any statement about using these medications together.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

Information from the Veterans Health Administration suggests low rates of pharmacotherapy for AUD. Approximately 3% of patients with AUD received a prescription for naltrexone, with less than 10% of those treated with naltrexone receiving long-acting injectable naltrexone (Iheanacho et al. 2013; Marienfeld et al. 2014).

Given the clinical considerations associated with the selection of a pharmacotherapy for a patient with AUD, a performance measure derived from this recommendation is not recommended. Clinical judgment would be needed to assess whether contraindications to treatment are present and to determine if there was a lack of response to nonpharmacological treatments alone. Increased documentation burden could result if each element of the recommendation needed to be recorded as standardized or structured text. Alternatively, if information was recorded as free text, additional time would be needed in reviewing documentation and determining if measure criteria were met. However, this recommendation could be used as a process-focused internal or health system–based quality improvement measure by tracking rates of prescribing for naltrexone and acamprosate in individuals with AUD. A quality measure could also examine receipt of AUD pharmacotherapy more broadly, as has been suggested by an expert panel convened by the RAND Corporation on AUD-related quality measures (Hepner et al. 2017). The American Society of Addiction Medicine (ASAM) Performance Measures for the Addiction Specialist Physician also include a suggested measure related to receipt of AUD pharmacotherapy (American Society of Addiction Medicine 2014). Changes in prescribing rates could be determined after initiatives to educate clinicians or reduce barriers to pharmacotherapy use (Abraham et al. 2011; Harris et al. 2016). Electronic decision support could identify individuals with a new diagnosis of moderate to severe AUD (as documented as a problem or diagnosis) and provide information on acamprosate and naltrexone for consideration by the clinician through a passive alert or “infobutton” (Del Fiol et al. 2012).

Statement 10: Disulfiram

• have a goal of achieving abstinence,

• prefer disulfiram or are intolerant to or have not responded to naltrexone and acamprosate,

• are capable of understanding the risks of alcohol consumption while taking disulfiram, and

• have no contraindications to the use of this medication.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

As a suggestion, this statement is inappropriate for use as a quality measure. However, a quality measure could examine receipt of AUD pharmacotherapy more broadly, as has been suggested by an expert panel convened by the RAND Corporation to identify potential AUD-related quality measures (Hepner et al. 2017). The ASAM Performance Measures for the Addiction Specialist Physician also include a suggested measure related to receipt of AUD pharmacotherapy (American Society of Addiction Medicine 2014).

Statement 11: Topiramate or Gababentin

• have a goal of reducing alcohol consumption or achieving abstinence,

• prefer topiramate or gabapentin or are intolerant to or have not responded to naltrexone and acamprosate, and

• have no contraindications to the use of these medications.

Implementation

Several additional medications may be efficacious in the treatment of moderate to severe AUD. These include topiramate and gabapentin. Although these medications will typically be used after trials of naltrexone and acamprosate, patient preference may lead to earlier use. Other factors that can guide medication selection include ease of administration, side effect profile, and the presence of co-occurring conditions that would affect treatment with a specific medication. There is no specific evidence on the optimal duration of treatment with these medications; such decisions are likely to be based on individual factors such as patient preference, disorder severity, history of relapses, potential consequences of relapse, clinical response, and tolerability. For discussion of the use of these medications in women who are pregnant or breastfeeding, see guideline Statement 14.

In clinical trials, topiramate was associated with significant reductions in the percent of heavy drinking days and the percent of drinking days in most (Johnson et al. 2003, 2007; Knapp et al. 2015; Kranzler et al. 2014a), but not all (Kampman et al. 2013; Likhitsathian et al. 2013), studies. Some studies also showed improvements in other drinking outcomes, such as drinks per drinking day and abstinence (Knapp et al. 2015; Kranzler et al. 2014a), the subjective experience of “craving” (Johnson et al. 2003; Kranzler et al. 2014b; Martinotti et al. 2014), and quality of life and well-being (Johnson et al. 2004a). Topiramate was typically administered at doses of 200–300 mg daily, but gradual dose titration may minimize some of the medication’s adverse effects. Because of its association with weight loss in 4%–21% of patients (Micromedex 2017d), topiramate may be a medication to consider in patients with obesity. Other common side effects of topiramate include sedation, cognitive dysfunction (e.g., effects on short-term memory, 3%–12%), dizziness (4%–25%), paresthesias (1%–51%), and gastrointestinal side effects (2%–11% vs. 6% in placebo) (Micromedex 2017d). In AUD patients, clinical trials most often reported dizziness, paresthesias, taste abnormalities, decreased appetite or weight loss, and cognitive or memory effects as occurring more often with topiramate than with placebo (Johnson et al. 2003, 2007; Kampman et al. 2013; Knapp et al. 2015; Kranzler et al. 2014a; Likhitsathian et al. 2013). Less common but notable side effects include metabolic acidosis, nephrolithiasis, and precipitation of acute angle-closure glaucoma. When initiating treatment with topiramate, it may be appropriate to assess cognitive status at baseline and renal function. In individuals with renal impairment, the dose of topiramate will need a reduction. Caution is also warranted in patients at risk for falls, including the elderly.

Gabapentin, at doses between 900 and 1800 mg/day, was associated with an increased rate of abstinence (number needed to treat [NNT] = 8 for 1800 mg daily) and a reduction in heavy drinking days (NNT = 5 for 1800 mg daily) in a 12-week double-blind, randomized, placebo-controlled, dose-ranging study (Mason et al. 2014). Reductions were also noted in drinking quantity and frequency, GGT, craving, mood, and insomnia. Several small randomized trials of shorter duration also showed benefit of gabapentin on alcohol-related outcomes (Anton et al. 2009, 2011; Furieri and Nakamura-Palacios 2007). In patients with AUD treated with gabapentin, there were no differences in the number, severity, or type of reported adverse effects (Mason et al. 2014). Fatigue (23%), insomnia (18%), and headache (14%) were noted most often, but rates with gabapentin did not differ from placebo. Some individuals have been noted to misuse gabapentin (Mersfelder and Nichols 2016), and attention for possible misuse may be warranted if prescribing gabapentin for treatment of AUD (Evoy et al. 2017). In individuals with renal impairment, the dose of gabapentin requires adjustment because gabapentin does not undergo metabolism and is excreted unchanged, predominantly in urine (Micromedex 2017b).

Other medications are being investigated for use in the treatment of AUD; however, the evidence for their use is more limited. Examples include zonisamide and ondansetron. Alcohol-related outcomes were also reduced by varenicline in several studies (de Bejczy et al. 2015; Litten et al. 2013), suggesting that varenicline may be a promising treatment of AUD, particularly for individuals with co-occurring nicotine dependence (Litten et al. 2016). Aripiprazole has been studied in a large multisite trial of AUD, and treatment was associated with reduced secondary endpoints of harm and reduced drinking, although the primary endpoint of abstinence did not differ from placebo (Anton et al. 2008a). However, aripiprazole could be considered in a patient with AUD and another indication for this medication. Findings with baclofen are mixed, and recent RCTs have found no benefit of its use in treatment for AUD (Beraha et al. 2016; Garbutt et al. 2010; Hauser et al. 2017; Ponizovsky et al. 2015). Nalmefene has also been studied in AUD in multiple European trials but is not available in the United States or Canada (Rösner et al. 2010).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

As a suggestion, this statement is inappropriate for use as a quality measure. However, a quality measure could also examine receipt of AUD pharmacotherapy more broadly, as has been suggested by an expert panel convened by the RAND Corporation on AUD-related quality measures (Hepner et al. 2017). The ASAM Performance Measures for the Addiction Specialist Physician also include a suggested measure related to receipt of AUD pharmacotherapy (ASAM 2014).

Implementation

Antidepressant medications are not recommended for treating AUD unless a co-occurring disorder such as a depressive or anxiety disorder is present that would warrant such treatment. When antidepressant medications have been studied in individuals with AUD and no co-occurring disorders, minimal efficacy was noted for alcohol-related outcomes for samples as a whole, and, in some studies, outcomes worsened. Individuals with specific patterns of high-risk or severe drinking or early onset of problem drinking may be more prone to exhibiting an increase in alcohol consumption (Dundon et al. 2004; Kranzler et al. 1996, 2012a; Pettinati et al. 2000), whereas a small subgroup of individuals with later onset and low risk or severity may show some benefit from antidepressants (Dundon et al. 2004; Kranzler et al. 1996, 2012a; Pettinati et al. 2000). Genetic factors (e.g., SLC6A4 genotype) may modulate these responses (Kranzler et al. 2011).

AUD often co-occurs with other psychiatric disorders, and individuals with AUD have increased odds of having a mood or anxiety disorder as compared with those without AUD (Lai et al. 2015). Among individuals with a 12-month prevalence of any AUD, the 12-month prevalence of any mood disorder is 18.9%, and the 12-month prevalence of any anxiety disorder (including specific phobia) is 17.1% (Grant et al. 2004). The 12-month prevalence of a mood or anxiety disorder is even larger in individuals with moderate to severe AUD and in individuals who seek treatment for AUD (Grant et al. 2004; Kaufmann et al. 2014). Thus, about one-third of individuals who seek treatment for AUD will have a major depressive disorder that is independent of alcohol intoxication or withdrawal (Grant et al. 2004). When AUD co-occurs with a mood disorder or an anxiety disorder, it can affect access to care and reduces treatment outcomes for both types of disorders (Drake et al. 2001). Consequently, the initial evaluation of a patient with AUD should include assessment for co-occurring psychiatric disorders. (See guideline Statement 4.) In determining whether an antidepressant medication is indicated for a co-occurring diagnosis, it is essential to keep in mind that use of alcohol or withdrawal from alcohol can be associated with mood or anxiety symptoms that can mimic a mood or anxiety disorder. In addition, AUD can be associated with psychosocial stressors (e.g., family issues, employment or financial difficulties, legal problems) that can also influence mood or be associated with anxiety. Thus, a careful consideration of differential diagnostic possibilities (American Psychiatric Association 2013) is important before embarking on treatment. Research is limited, but with depression, symptoms that appear to be independent of alcohol may respond better to antidepressant medications than do symptoms of substance-induced depression (Foulds et al. 2015). If an antidepressant medication is indicated for a co-occurring condition, it can be used in combination with other AUD pharmacotherapies. For example, in one randomized placebo-controlled trial conducted in individuals with AUD and major depressive disorder, a combination of sertraline plus naltrexone produced a higher rate of abstinence and a longer time to relapse to heavy drinking than naltrexone alone, sertraline alone, or placebo, with fewer serious adverse events with combination treatment as compared with other treatment groups (Pettinati et al. 2010). In another study of predominantly male veterans with co-occurring AUD and PTSD, all groups showed reductions in PTSD symptoms, but combination treatment with naltrexone and an antidepressant was associated with reduced craving as compared with groups treated with antidepressant plus placebo (Petrakis et al. 2012).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure because the recommendation would not pertain to the majority of individuals with AUD. However, this recommendation may be appropriate for use in the Choosing Wisely initiative. It could also be used as an internal or health system–based quality improvement measure if prescribing of antidepressant medications appears to be frequent among patients with AUD. Furthermore, this recommendation could be integrated into electronic clinical decision support. If an order for an antidepressant is entered for an individual with AUD, the clinicians could be alerted to consider whether or not antidepressant therapy is indicated. The alert could be configured so that it would not be presented to the clinician for patients with a documented problem or diagnosis of major depressive disorder, obsessive-compulsive disorder, PTSD, or panic disorder with or without agoraphobia.

Statement 13: Benzodiazepines

Implementation

There is no evidence for the use of benzodiazepines in the primary treatment of AUD, except for alcohol detoxification or the treatment of alcohol withdrawal, which are outside the scope of this practice guideline. Similarly, there is no evidence related to the use of other sedative-hypnotic medications such as barbiturates, meprobamate, or nonbenzodiazepine hypnotics (zolpidem, eszopiclone, zaleplon) in individuals with AUD. Clinicians should exercise caution because the use of benzodiazepines or other sedative-hypnotic agents in the setting of alcohol intoxication carries with it an increased risk for sedation, behavioral impairment, respiratory depression, and death in severe cases (Bachhuber et al. 2016). Clinicians should discuss this risk with patients who are actively drinking alcohol and consider other treatments or medications when possible. For example, in a patient with an anxiety disorder and AUD, use of psychotherapy or an antidepressant medication would be indicated before considering a benzodiazepine or other sedating medication with addictive potential. Although there may still be limited circumstances in which prescribing a benzodiazepine or sedative-hypnotic agent is appropriate for treating a co-occurring disorder, if a benzodiazepine is prescribed, one might consider prescribing only a limited quantity at the lowest possible dose in order to mitigate potential risks.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure. Most clinicians are already aware of the potential difficulties in using benzodiazepines to treat an individual with AUD unless acute alcohol withdrawal or another appropriate indication is present. However, this recommendation may be appropriate for use in the Choosing Wisely initiative. In addition, this recommendation may be appropriate for integration into electronic clinical decision support. Clinicians could be alerted to consider whether an appropriate indication exists for benzodiazepine treatment if a benzodiazepine order is entered for an individual with a documented problem with or diagnosis of AUD.

Statement 14: Pharmacotherapy in Pregnant or Breastfeeding Women

Implementation

Alcohol use during pregnancy can contribute to an increased risk of congenital malformations or later intellectual disability (Harris et al. 2017; Huang et al. 2016) as well as an increased risk of fetal alcohol spectrum disorders (Gupta et al. 2016; Riley et al. 2011). With pharmacotherapies for AUD, there is limited evidence regarding the potential risks to an exposed fetus or infant (Briggs and Freeman 2015). As with other medications, however, the risks to the fetus are likely to be greatest during the first trimester (Mitchell et al. 2011). On the basis of data in pregnant women, there does appear to be an increased risk of malformation associated with the use of topiramate (Alsaad et al. 2015; Briggs and Freeman 2015; Tennis et al. 2015; Weston et al. 2016). Data in pregnant animals are not available for disulfiram but suggest a moderate risk for use of naltrexone, high risk for use of acamprosate, and possible risks for use of gabapentin and topiramate (Briggs and Freeman 2015). For these reasons, it is recommended that nonpharmacological interventions be used preferentially for treating AUD during pregnancy. For individuals who become pregnant while taking a medication to treat AUD, the risk of continuing or stopping pharmacological treatment should be individualized to the patient and discussed with the patient, her obstetrician, and, if applicable, her partner. Potential risk to the fetus from medication should be balanced against the risk of relapse to alcohol use, which itself carries teratogenic risk.

Decisions about breastfeeding and use of these medications in breastfeeding women also require individualized discussion with the patient and the infant’s pediatrician in balancing the benefits of breastfeeding and potential harms of exposure to medication in breast milk. Again, data are limited, but there may be potential for toxicity with disulfiram, naltrexone, and topiramate (Briggs and Freeman 2015), whereas acamprosate and gabapentin are noted to be “probably compatible” with breast-feeding (Briggs and Freeman 2015).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure. The recommendation would not pertain to the majority of individuals with AUD, and adherence with this recommendation is already likely to be high as a result of the patient and clinician concern about use of medication while the patient is pregnant or breastfeeding. However, this recommendation may be appropriate for integration into electronic clinical decision support. In women who are pregnant or breastfeeding, clinicians could be alerted to avoid pharmacotherapy for AUD except under the circumstances noted in the recommendation.

Statement 15: Acamprosate in Severe Renal Impairment

Implementation

Because of the excretion of acamprosate through the kidneys, serum creatinine should be measured at baseline, and in individuals with a history of renal impairment, results should be reviewed before initiating treatment. A CrCl less than 30 mL/min or an eGFR less than 30 mL/min/1.73 m² is a contraindication to the use of acamprosate, and a different medication such as naltrexone should be used.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure. Adherence to this recommendation is already likely to be high as a result of the FDA warning about use of acamprosate in individuals with severe renal impairment. However, this recommendation may be appropriate for integration into electronic clinical decision support. Clinicians could be alerted to use a different pharmacotherapy for AUD in individuals with a documented problem or diagnosis of severe renal impairment.

Statement 16: Acamprosate in Mild to Moderate Renal Impairment

Implementation

Because of the excretion of acamprosate through the kidneys, serum creatinine should be measured at baseline, and in individuals with a history of renal impairment, results should be reviewed before initiating treatment. For a CrCl between 30 and 50 mL/min or eGFR between 30 and 59 mL/min/1.73 m², a reduced dose of 333 mg three times per day is suggested. Alternatively, a different medication such as naltrexone could be used.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure. Although clinicians may be less aware of the need to adjust the dosing of acamprosate in mild to moderate renal impairment, the recommendation would not pertain to the majority of individuals with AUD. However, this recommendation may be appropriate for integration into electronic clinical decision support. Clinicians could be alerted to consider a different pharmacotherapy for AUD in individuals with a documented problem or diagnosis of renal impairment. If an order for acamprosate is placed after review of the preceding alert, clinical decision support could advise adjusting the dose of the medication in proportion to the degree of renal impairment.

Statement 17: Naltrexone in Acute Hepatitis or Hepatic Failure

Implementation

On the basis of data from clinical trials, a fraction of individuals treated with naltrexone exhibit increases in hepatic enzyme levels or other signs of hepatocellular injury (Anton et al. 2006). However, other studies have suggested that rates of elevated hepatic enzymes with naltrexone are comparable to rates with placebo, even in patient populations at increased risk for hepatic dysfunction due to co-occurring hepatitis C or HIV infection (Croop et al. 1997; Lucey et al. 2008; M. C. Mitchell et al. 2012; Tetrault et al. 2012; Vagenas et al. 2014). Nevertheless, it seems prudent to avoid exposure to naltrexone in individuals who are already experiencing significant evidence of liver damage such as acute hepatitis or hepatic failure. Acamprosate could be considered in these individuals because of its lack of hepatic effects.

Liver chemistries are appropriate to obtain before treating a patient with naltrexone, with additional evaluation or consultation and follow-up liver chemistries, as indicated, depending on the extent of any abnormalities (Kwo et al. 2017). In making a determination about use of naltrexone in an individual with some elevations in liver chemistries, it is important to recognize that individuals with AUD will often have hepatic enzyme abnormalities due to alcohol use, other medications, infectious etiologies (e.g., hepatitis C), or obesity. In clinical trials, individuals were generally not excluded unless hepatic enzyme levels were more than 3 times the upper limit of normal. For comparison, the case definition of acute hepatitis C includes a peak elevated serum ALT level > 200 IU/L (Centers for Disease Control and Prevention 2016), and the definition of drug-induced hepatitis includes peak ALT levels above 800 IU/L, or about 20 times the upper limit of normal (National Library of Medicine 2017b). Furthermore, decisions about whether or not to use naltrexone will balance an infrequent but potentially negative effect of naltrexone on liver function with the improvements in liver chemistries that are associated with successful treatment of AUD.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure. Adherence to this recommendation is already likely to be high as a result of the FDA warning about use of naltrexone in individuals with acute hepatitis or hepatic failure. However, this recommendation may be appropriate for integration into electronic clinical decision support. Clinicians could be alerted to consider a different pharmacotherapy for AUD in individuals with a documented problem with or diagnosis of acute hepatitis or hepatic failure.

Statement 18: Naltrexone With Concomitant Opioid Use

Implementation

Because naltrexone is an opioid receptor antagonist, it is efficacious in treating both AUD and opioid use disorder. However, before starting naltrexone in either its oral or long-acting injectable formulation, outpatients must be abstinent from opioids for 7–14 days (depending on the duration of action of the opioid) because of the risk for precipitating opioid withdrawal. It is also important that patients understand the risk of precipitated withdrawal if they continue to use opioids during treatment initiation with naltrexone. Strategies for minimizing the risk of opioid withdrawal might include starting with a small test dose of oral naltrexone (e.g., 25 mg) and/or obtaining a urine drug screen for opioids before initiating treatment.

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure because among individuals who present for treatment of AUD, the fraction of patients who use or have an anticipated need for opioids is likely to be small. However, this recommendation may be appropriate for integration into electronic clinical decision support. At the time of placing an initial order for naltrexone, clinicians could be alerted to consider whether the individual is currently using opioids or has an anticipated need for opioids.

Implementation

Because naltrexone is an opioid receptor antagonist, it is efficacious in treating both AUD and opioid use disorder; however, it can be considered only in individuals who wish to abstain from opioid use. Methadone and buprenorphine exhibit the best efficacy for treatment of opioid use disorder (Kampman and Jarvis 2015; Mattick et al. 2014), and acamprosate or other pharmacotherapies can be given concomitantly to address AUD. If naltrexone is used for individuals with co-occurring AUD and opioid use disorder, adherence can be assured more readily by the use of a long-acting formulation of naltrexone. Studies of oral naltrexone in individuals with opioid use disorder suggest that with poor adherence, efficacy is limited and mortality is greater than with long-acting injectable or implantable preparations (Degenhardt et al. 2015; Kelty and Hulse 2012; Krupitsky et al. 2011, 2012). (See guideline Statement 9 for additional details on treatment of patients with naltrexone.)

Before starting naltrexone, outpatients must be abstinent from opioids for 7–14 days (depending on the duration of action of the opioid) because of the risk of precipitating opioid withdrawal. For patients being treated with an opioid agonist (e.g., methadone, buprenorphine), protocols have been developed for transitioning to antagonist therapy with naltrexone (Mannelli et al. 2012). In addition, strategies for minimizing the risk of opioid withdrawal might include starting with a small test dose of oral naltrexone (e.g., 25 mg) and/or obtaining a urine drug screen for opioids before initiating treatment. If prescription-related information is available through an electronic medical record or prescription drug monitoring program, it should be checked for current or recent opioid prescriptions. Coordinating care with other clinicians is also important. Patients should be warned about returning to opioid use after being on naltrexone because tolerance will be much reduced and the risk of a fatal overdose is greater. It is also advisable for patients to carry a wallet card noting that they are taking naltrexone so this information will be available to emergency personnel. A template for wallet cards as well as sample templates for documenting medication management visits are available through NIAAA (U.S. Department of Health and Human Services 2007).

Balancing of Potential Benefits and Harms in Rating the Strength of the Guideline Statement

Quality Measurement Considerations

This statement is not likely to be appropriate for use as a quality measure because the fraction of patients who have AUD and a co-occurring opioid use disorder is likely to be small. However, this recommendation may be appropriate for integration into electronic clinical decision support. Clinicians could be alerted to consider whether naltrexone would be an appropriate pharmacotherapy for individuals with documented AUD and opioid use disorder as a problem or diagnosis.