A nurse is caring for a client who takes naltrexone for the treatment of alcohol use disorder

JAMA Intern Med. 2018 May; 178(5): 613–621.

A Randomized Clinical Trial

Katharine A. Bradley, MD,

Key Points

Question

Does an offer of individualized alcohol care management from a nurse for up to 12 months—in which patients choose their drinking goals and alcohol-related care—improve drinking outcomes for primary care patients with heavy drinking?

Findings

In this randomized clinical trial of 304 participants, patient-centered alcohol care management did not decrease heavy drinking or alcohol-related problems at 12 months even though more patients engaged in alcohol-related care, including medications for alcohol use disorders.

Meaning

Additional research is needed to develop effective primary care management for patients at high risk for alcohol use disorders.

Abstract

Importance

Experts recommend that alcohol use disorders (AUDs) be managed in primary care, but effective approaches are unclear.

Objective

To test whether 12 months of alcohol care management, compared with usual care, improved drinking outcomes among patients with or at high risk for AUDs.

Design, Setting, and Participants

This randomized clinical trial was conducted at 3 Veterans Affairs (VA) primary care clinics. Between October 11, 2011, and September 30, 2014, the study enrolled 304 outpatients who reported heavy drinking (≥4 drinks per day for women and ≥5 drinks per day for men).

Interventions

Nurse care managers offered outreach and engagement, repeated brief counseling using motivational interviewing and shared decision making about treatment options, and nurse practitioner–prescribed AUD medications (if desired), supported by an interdisciplinary team (CHOICE intervention). The comparison was usual primary care.

Main Outcomes and Measures

Primary outcomes, assessed by blinded telephone interviewers at 12 months, were percentage of heavy drinking days in the prior 28 days measured by timeline follow-back interviews and a binary good drinking outcome, defined as abstinence or drinking below recommended limits in the prior 28 days (according to timeline follow-back interviews) and no alcohol-related symptoms in the past 3 months as measured by the Short Inventory of Problems.

Results

Of 304 participants, 275 (90%) were male, 206 (68%) were white, and the mean (SD) age was 51.4 (13.8) years. At baseline, both the CHOICE intervention (n = 150) and usual care (n = 154) groups reported heavy drinking on 61% of days (95% CI, 56%-66%). During the 12-month intervention, 137 of 150 patients in the intervention group (91%) had at least 1 nurse visit, and 77 of 150 (51%) had at least 6 nurse visits. A greater proportion of patients in the intervention group than in the usual care group received alcohol-related care: 42% (95% CI, 35%-49%; 63 of 150 patients) vs 26% (95% CI, 19%-35%; 40 of 154 patients). Alcohol-related care included more AUD medication use: 32% (95% CI, 26%-39%; 48 of 150 patients in the intervention group) vs 8% (95% CI, 5%-13%; 13 of 154 patients in the usual care group). No significant differences in primary outcomes were observed at 12 months between patients in both groups. The percentages of heavy drinking days were 39% (95% CI, 32%-47%) and 35% (95% CI, 28%-42%), and the percentages of patients with a good drinking outcome were 15% (95% CI, 9%-22%; 18 of 124 patients) and 20% (95 % CI, 14%-28%; 27 of 134 patients), in the intervention and usual care groups, respectively (P = .32-.44). Findings at 3 months were similar.

Conclusions and Relevance

The CHOICE intervention did not decrease heavy drinking or related problems despite increased engagement in alcohol-related care.

Trial Registration

clinicaltrials.gov Identifier: NCT01400581

Introduction

More than 13% of US adults have active alcohol use disorders (AUDs),1,2 and fewer than 1 in 5 receive alcohol treatment.1,3 Even when referred, most people with AUDs do not engage in alcohol treatment.4,5 Moreover, AUDs are often chronic.6,7,8,9,10,11

Experts have called for new models of primary care (PC) to manage AUDs.6,7,8,9,10,11 These models focus on engaging patients in patient-centered PC, often with shared decision making, because no one treatment among a number of effective medications and behavioral treatments for AUDs is consistently superior.12,13,14,15 As a result, engagement in alcohol-related care has been identified as a key measure of quality of care for AUDs, including in PC.16,17 However, trials testing PC management of AUDs have had varying results.4,18,19,20

The Choosing Healthier Drinking Options in Primary Care (CHOICE) intervention was designed to improve drinking outcomes by engaging patients at high risk for AUDs in patient-centered, alcohol-related care, including individualized drinking goals.21 The CHOICE trial tested whether the 12-month CHOICE intervention decreased the frequency of heavy drinking and increased the proportion of patients drinking below recommended limits without alcohol-related symptoms.

Methods

Study Design and Setting

The CHOICE trial was a randomized encouragement trial22 that offered patients alcohol-related services but did not require that they accept.21 The trial was conducted at 3 Veterans Affairs (VA) PC sites in Washington (1 added after the trial commenced to meet recruitment targets) and was approved with Health Insurance Portability and Accountability Act waivers by VA Puget Sound and Kaiser Permanente Washington institutional review boards. Enrollment coordinators recruited patients and obtained written informed consent. The study had a Certificate of Confidentiality from the National Institutes of Health. The full trial protocol is available in Supplement 1.

Eligibility and Recruitment

Primary care patients were eligible for recruitment if they had positive screening results for AUDs in their VA electronic health records (EHRs) based on Alcohol Use Disorders Identification Test–Consumption (AUDIT-C) scores (≥4 points for women and ≥5 for men).23 The multistep recruitment process included asking PC providers (eg, physicians, nurse practitioners, and physician assistants) to review their patients with positive screening results for appropriateness and to sign study invitation letters if willing; sending patients invitation letters with an opt-out option; telephone screening; and in-person baseline assessments.21 About 25% of men eligible for recruitment were randomly assigned to a no-contact group to allow evaluation of recruitment effects.21

Inclusion criteria,21 designed to identify patients who might benefit from alcohol care management, included receipt of PC in a study clinic, age 21 to 75 years, and heavy drinking (≥4 drinks/occasion for women and ≥5 for men,24 at least twice per week in the past month or once per week if prior alcohol treatment).25 Exclusions were acute medical or psychiatric instability, alcohol treatment in the previous 90 days, cognitive impairment, end-of-life care, current or planned pregnancy, enrollment in another VA trial, plans to leave the VA, employment by the VA, exclusion by PC provider, or inadequate contact information.

Randomization

Patients were randomized 1:1 to usual VA PC alone or usual care plus the intervention in blocks of 6, 8, and 10, stratified by sex, past-year Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition alcohol dependence, and site using a computer-generated list of random numbers. Treatment assignment was concealed by computer and assigned by a study coordinator after baseline assessments were complete.21 Primary care providers were given a general description of the intervention, and randomization assignments were noted in the EHR.

CHOICE Intervention

Two nurses provided alcohol care management over 12 months, consistent with effective approaches for other chronic conditions,26,27 communicating with PC providers via the VA EHR.21 A PC nurse practitioner was available to prescribe US Food and Drug Administration–approved medications for AUDs,14 and an interdisciplinary team supervised nurses weekly. Nurses received 6 to 18 hours of training in motivational interviewing and AUDs, depending on prior experience.21

Nurses initially invited patients to 1 to 2 engagement visits focused on the patients’ life and goals. Subsequently, nurses provided feedback from baseline assessments, using motivational interviewing and shared decision making28 to engage patients in considering options regarding drinking (no change, self-monitoring, decreasing drinking, or abstaining), and asked if patients wanted support for making changes. Support included repeated nurse visits to review patient self-monitoring and/or biomarkers and provide behavioral goal setting and skills development for reducing drinking, AUD medications,14 withdrawal management, mutual help (eg, Alcoholics Anonymous), and referral to and/or support for linkage to VA specialty addictions treatment, per patient preference. The frequency of nurse visits and whether they occurred by telephone or in person were also based on patient preference. However, nurses recommended follow-up every 1 to 2 weeks for 2 months and monthly thereafter, with the next appointment scheduled at the end of each contact.21 Patients who missed appointments were called weekly, then monthly, and sent letters.

Usual Care

All patients received usual care from their VA PC providers. The 3 study sites all offered annual behavioral health screening, integrated mental health services, and access to specialty mental health and addictions clinics.

Data Collection

Baseline assessments included in-person interviews, self-report questionnaires, and laboratory tests. Drinking outcomes were assessed at 3 months and 12 months by independent, blinded telephone interviewers.29 Laboratory biomarkers and self-report questionnaires were also collected at 12 months.21 Data on utilization and medications were obtained from VA databases. Patients were compensated $20, $10, and $30 for baseline, 3-month, and 12-month assessments, respectively.

Measures

Primary Outcomes

Two primary outcomes, measured at 12 months, were percentage of heavy drinking days (HDDs)30 and good drinking outcomes (GDOs).31,32 These outcomes were relevant irrespective of whether a patient’s goal was abstinence. Percentage of HDDs was the proportion of nonhospitalized days when the patient reported heavy drinking (defined earlier) on the 28-day timeline follow-back interview.33 Percentage of HDDs is associated with AUD symptoms.30,34 A GDO was defined as abstinence or drinking below recommended limits based on timeline follow-back interviews (0 HDDs and ≤7 or ≤14 drinks per week for women and men, respectively), and no alcohol symptoms (past 3 months) on the Short Inventory of Problems (SIP).35

Secondary Outcomes

Process measures of engagement in alcohol-related care relevant to patients in both the intervention and usual care groups were as follows: receipt of AUD medications (naltrexone, acamprosate, or disulfiram); greater than 30 days’ supply of AUD medication(s); VA addictions treatment; patient-reported Alcoholics Anonymous attendance; and a composite of any of these. Measures of participation in the intervention included the number of visits with a CHOICE nurse and receipt of abnormal alcohol biomarker monitoring on 2 or more occasions (gamma glutamyl transferase, mean corpuscular volume, or carbohydrate deficient transferrin). Secondary drinking outcomes, defined a priori, included percentage of HDDs and GDO at 3 months and others measured at 3 and 12 months: no heavy drinking, percentage of days abstinent, abstinence, drinking below weekly limits (based on timeline follow-back interviews), SIP scores (past 3 months),35 and readiness rulers.36 At 12 months, patients also completed the AUDIT (0-40 points),37 including the AUDIT-C (0-12 points),23 and testing for alcohol biomarkers. Measures of health status (1-5 points)38 and hospitalizations were assessed at both follow-ups. Health care utilization and remaining covariates (Table 1)39,40,41,42 were obtained from baseline assessments or EHRs. Covariates for main analyses included stratification variables (sex, alcohol dependence, site), baseline age, and baseline value of the outcome.21

Table 1.

Baseline Characteristics of Randomized Participants

Sample Size

A sample size of 150 patients per group was needed to detect a difference of 4 HDDs (out of 28), based on a type I error probability of 0.05 (2-sided), to provide greater than 90% power, assuming attrition of less than 15% and an intraclass correlation coefficient of 0.05 to account for clustering of patients by PC providers, so recruitment continued until 300 patients consented. This sample size provided greater than 80% power to detect a 16% absolute difference in the proportion of patients with GDOs.

Statistical Analysis

Analyses comparing the intervention and usual care groups followed an intent-to-treat approach, with patients included in randomly assigned groups regardless of participation. In accordance with the prespecified analysis plan, differences in primary and secondary outcomes between intervention and usual care groups were evaluated using generalized estimating equations regression models clustered by PC provider, adjusted for covariates described earlier.43 For the binary GDO, eligibility criteria required baseline values of 0, so analyses were adjusted for baseline percentage of HDDs. Analyses used an independent working correlation model and a robust sandwich variance estimator. For the primary outcomes, percentage of HDDs, which was nonnormally distributed (eFigure in Supplement 2), and GDO, binomial regression with a logit link was used to estimate odds ratios. For the secondary outcome percentage of days abstinent and binary outcomes (eg, abstinence), the same approaches were used. For continuous secondary outcomes (eg, laboratory tests), linear regression models were used to estimate mean differences. For count data (eg, number of PC visits), Poisson models with log link were used to estimate relative risks. For health-related quality of life, proportional odds models with a logit link were used to estimate odds ratios. Tests and confidence limits were 2-sided (α = .05).44

To account for missing follow-up data, primary analyses applied multiple imputation using chained equations,45 which assumes data are missing at random,46 based on 30 imputations (eAppendix 1 in Supplement 2). Sensitivity analyses were conducted including only complete cases, as planned a priori.

Recruitment alone could decrease drinking by patients in the usual care group,47,48,49 biasing the trial to the null. Therefore, the study included a no-contact group to allow evaluation of recruitment effects.21 Scores on the AUDIT-C (0-12 points) documented in the EHR in the 24 months after EHR sampling21 were compared for potentially recruited men and those randomly assigned to the no contact group (Figure), using generalized estimating equations adjusting for baseline EHR AUDIT-C scores and time to follow-up EHR AUDIT-C scores (eAppendix 2 in Supplement 2). Analyses were conducted using the Microsoft R Open software package, version 3.3.1 (Microsoft Corp).

Screening, Recruitment, and Follow-up of Study Sample

Primary care patients were eligible for recruitment if they had positive screening results for alcohol use disorders in their Veterans Affairs electronic health records. The recruitment process included asking primary care providers to review patients with positive screening results for appropriateness and sign study invitation letters if willing; sending patients invitation letters with an opt-out option; telephone screening; and in-person baseline assessments. The no-contact group was used to evaluate potential recruitment effects. EHR indicates electronic health record.

Results

Recruitment and Follow-up

Of 1400 patients who completed telephone screening, 571 were eligible, and 304 (53%) consented and were randomized (Figure). Two hundred seventy-five participants (90%) were male, 206 (68%) were white, and the mean (SD) age was 51.4 (13.8) years. The intervention (n = 150) and usual care (n = 154) groups were comparable at baseline (Table 1), and follow-up was comparable (Figure; eTable 1 in Supplement 2). Two hundred fifty-eight patients (85%) completed interviews at 12 months.

Receipt of the Intervention and Alcohol-Related Care

Among participants randomized to the CHOICE intervention, 137 of 150 (91%) had at least 1 visit with a CHOICE nurse and 77 of 150 (51%) had at least 6 visits (8 [5%] had 1 visit; 52 [35%], 2-5 visits; 32 [21%], 6-9 visits; and 45 [30%], ≥10 visits). While few patients in either group had biomarker monitoring, more in the intervention group had it than in usual care (17 of 150 [11%] vs 3 of 154 [2%]; odds ratio, 6.8; 95% CI, 1.8-25.9; P = .005). More patients in the intervention group than in the usual care group were treated with AUD medications (respective number of patients: oral naltrexone, 39 and 11; acamprosate, 12 and 0; and disulfiram, 4 and 5), but there was no difference in VA addiction treatment at 3 or 12 months (Table 2).

Table 2.

Alcohol-Related Care Received by Patients in Usual Care and Intervention Groups

Primary Outcomes

Primary outcomes improved at 12 months but did not differ between the groups. The mean percentage of HDDs decreased from 61% (95% CI, 56%-66%) at baseline to 39% (95% CI, 32%-47%) and 35% (95% CI, 28%-42%) in the intervention and usual care groups at 12 months, respectively (Table 3) (P = .44). The percentages of patients with GDOs were 15% (95% CI, 9%-22% [18 of 124 patients]) and 20% (95% CI, 14%-28% [27 of 134 patients]) in the intervention and usual care groups, respectively (P = .32) (Table 3). Results were unchanged in sensitivity analyses restricted to complete cases (eTable 2 and eTable 3 in Supplement 2).

Table 3.

Primary Outcomes at 12 Months and Secondary Outcomes at 3 and 12 Months

Secondary Outcomes

Secondary drinking outcomes at 3 and 12 months showed no benefit of the intervention (Table 3). Percentage of days abstinent was higher in patients in the usual care group at 3 and 12 months, whereas health status showed greater improvement in patients in the intervention group at 12 months (Table 3). Outpatient utilization did not differ (eTable 4 in Supplement 2) except specialty mental health care utilization was lower for patients in the intervention group at 12 months. The remainder of secondary outcomes, including biomarkers, did not differ between groups (Table 3 and eTable 4 in Supplement 2).

Analyses of the no-contact group suggested that recruitment did not decrease patients’ drinking. Mean adjusted EHR AUDIT-C scores during follow-up did not differ between potentially recruited men and those in the no contact group (mean [SD] score of 4.2 [0.09] in the no-contact group vs 4.3 [0.05] in the potentially recruited group; P = .40) (Table 4).21

Table 4.

Comparison of Alcohol Use Disorders Identification Test–Consumption Scores From Patients’ Electronic Health Records During Follow-up in No-Contact and Potentially Recruited Groups of Men

Discussion

The CHOICE trial tested whether offering alcohol care management to PC patients with or at high risk for AUDs improved drinking outcomes. The intervention did not decrease drinking or related symptoms despite increasing engagement in alcohol-related care, including a 4-fold increase in initiation of AUD medications. The intervention was associated with fewer days abstinent than usual care.

Four previous trials of alcohol care management showed increased patient engagement in alcohol-related care, but findings regarding changes in patients’ drinking have been mixed. One VA trial of PC patients4 tested an intervention that included a recommendation to take naltrexone and abstain from drinking, with frequent visits to encourage both; 66% of patients receiving the intervention took naltrexone and the intervention decreased heavy drinking (secondary outcome). Another trial recruited hospitalized VA patients and tested an intervention focused on PC management of alcohol-related medical conditions with a goal of abstinence.18 The intervention improved 3 of 4 drinking outcomes.18 The AHEAD trial recruited individuals with AUDs (77%) and/or drug use disorders (DUDs) (88%) from outside PC (>74% from residential detoxification) and evaluated PC delivered by an interdisciplinary team including a care manager.19 The patient-centered intervention increased AUD medication use from 4% to 16%, but did not improve primary or secondary drinking outcomes. The SUMMIT trial tested care management in PC patients with a high probability of AUDs and/or opioid use disorders. Care coordinators offered referral for brief therapy in PC and/or to PC providers for medications. The intervention did not increase AUD or opioid use disorder medication use (both 13%), but increased abstinence from both alcohol and opioids, the primary outcome: 33% vs 22% were abstinent at follow-up (P = .03).20

Differing results between the 2 alcohol care management trials with negative findings, CHOICE and AHEAD,19 and the other 3 trials that found that care management reduced drinking4,18,20 could reflect several differences across trials. One difference was the study samples. The trials in which care management resulted in decreased drinking were either restricted to patients with AUDs only (ie, no other DUDs) or showed benefit restricted to those with AUDs only.4,18,20,50 Care management interventions might be more effective in patients with uncomplicated AUDs. However, fewer than 1 in 5 patients in CHOICE had a DUD, making it unlikely that this is the sole factor contributing to observed differences in outcomes across trials. Severity of alcohol-related symptoms does not seem to distinguish positive and negative trials. At baseline, mean SIP scores were 9 in CHOICE, 20 to 21 for alcohol in AHEAD, 6 in the VA trial, and 10 to 11 in SUMMIT.4,19,20

Other aspects of trial design could have contributed to differing findings. The CHOICE and AHEAD trials, in which care management failed to reduce drinking, had prespecified primary drinking outcomes and rigorously assessed bias due to missing follow-up data.19 The other 3 trials did not.4,18,20 However, several factors make it unlikely that improved drinking outcomes in the latter trials were spurious findings.4 One trial showed decreased heavy drinking, a single prespecified secondary outcome.4 Another found that 3 of 4 drinking outcomes significantly improved.18 The third had a composite outcome of abstinence from both opioids and alcohol, but secondary analyses showed the benefit was restricted to patients with AUDs alone.20,50 The SUMMIT trial had lower follow-up (66%) that differed somewhat across the groups (62% and 70%) and could have impacted results. The CHOICE and AHEAD trials found results unchanged with multiple imputation and complete case analyses of primary outcomes; however, follow-up rates were much higher.

Specific elements of the alcohol care management interventions seem likely contributors to the different outcomes of alcohol care management trials. The trials that showed decreased drinking in the intervention compared with the control condition were all abstinence oriented. Two of the 3 tested more directive, yet patient-centered, interventions in which patients were presented with a treatment plan with an explicit recommendation to abstain.4,18 The SUMMIT trial did not specify that the intervention was abstinence oriented, but the main outcome was abstinence. The SUMMIT intervention also offered patients a manual-driven behavioral treatment in PC.20 In contrast, CHOICE and AHEAD allowed patients to select their own drinking goals, abstinence or reduction, based on an assumption that engaging patients in alcohol-related care would decrease drinking.19 In sum, the CHOICE trial’s results, when added to existing literature,4,18,19,20 suggest that to be effective, alcohol care managers might need to recommend abstinence, engage most patients in use of naltrexone,4 and/or offer effective behavioral treatment in PC.20

Two trials of alcohol care management with prespecified primary drinking outcomes—CHOICE and AHEAD19—found that the experimental interventions increased engagement in alcohol-related care without improving drinking outcomes. Current quality measures for AUDs are based on the assumption that engagement in alcohol-related care alone improves outcomes. The National Committee for Quality Assurance (NCQA) Healthcare Effectiveness Data and Information Set (HEDIS) AUD measures are focused on initiation and engagement in alcohol-related care. Any visit with an eligible AUD diagnosis code is considered initiation or engagement, even a PC visit without any evidence of medication or behavioral treatment for AUDs.51 The CHOICE and AHEAD trials suggest that engagement in alcohol-related PC management alone may be insufficient to improve drinking outcomes in patients with or at high risk for AUDs. This finding has implications for future improvements for quality measurement.

Limitations

The CHOICE trial has important limitations. First, while the intervention was designed to address AUDs, only 73% of participants met criteria for AUDs at baseline. Second, the multistep recruitment could have motivated some patients receiving usual care to change and/or seek care,47,48,49 and EHR documentation of randomization to the usual care group could have led clinicians caring for patients in the usual care group to refer them for alcohol treatment. More than 25% participated in specialty treatment or Alcoholics Anonymous by 12 months, and patients in the usual care group reported more days abstinent at both follow-ups. Future practical clinical trials should be designed to evaluate PC interventions compared with true usual care without recruitment, screening, and assessment of the comparison group until follow-up.52 Third, CHOICE was an effectiveness trial and did not assess intervention fidelity because of concern that audiotapes would interfere with engagement.21 Despite 2 hours of supervision weekly,21 the CHOICE nurses might not have offered the intervention as intended. Fourth, main measures were based on potentially biased self-report and the intervention could have made patients more comfortable accurately reporting their drinking. However, objective biomarkers did not differ between the groups. Fifth, the CHOICE trial’s main outcomes, commonly used in trials, may not reflect patient goals and those who “fail” on such outcomes could nonetheless have improved functioning.53,54 Future analyses will evaluate mental health symptoms and health status in multiple domains. Finally, only 53% of eligible patients enrolled, and the study sample consisting largely of middle-aged, male US veterans may limit generalizability.

Conclusions

The CHOICE intervention engaged patients in alcohol-related care and increased use of AUD medications. However, the nurse care management intervention did not decrease drinking more than usual care. Patient-centered interventions that recommend a goal of abstinence and engage most patients in medication use and/or offer brief therapy for AUDs appear most effective. Future research should evaluate whether these interventions improve functioning and health outcomes, as well as patient-reported drinking outcomes.

Notes

Supplement 1.

Trial Protocol

Supplement 2.

eFigure. Distribution of Heavy Drinking Days in the Past 28 days Outcome at Baseline, 3 Months and 12 Months

eAppendix 1. Description of approach for handling missing data

eAppendix 2. Methods for Comparing Follow-up EHR AUDIT-Cs in the “No Contact” and “Potentially Recruited” Groups of Men

eTable 1. Summary of missing data

eTable 2. Association of baseline covariates with missingness of primary outcomes

eTable 3. Secondary Outcomes at Three and Twelve Months

eTable 4. Comparison of treatment effect estimates from multiple imputation to those from complete case analysis

eReferences

References

1. Grant BF, Goldstein RB, Saha TD, et al.. Epidemiology of DSM-5 Alcohol Use Disorder: Results From the National Epidemiologic Survey on Alcohol and Related Conditions III. JAMA Psychiatry. 2015;72(8):757-766. [PMC free article] [PubMed] [Google Scholar]

2. Rehm J, Dawson D, Frick U, et al.. Burden of disease associated with alcohol use disorders in the United States. Alcohol Clin Exp Res. 2014;38(4):1068-1077. [PMC free article] [PubMed] [Google Scholar]

3. McGlynn EA, Asch SM, Adams J, et al.. The quality of health care delivered to adults in the United States. N Engl J Med. 2003;348(26):2635-2645. [PubMed] [Google Scholar]

4. Oslin DW, Lynch KG, Maisto SA, et al.. A randomized clinical trial of alcohol care management delivered in Department of Veterans Affairs primary care clinics versus specialty addiction treatment. J Gen Intern Med. 2014;29(1):162-168. [PMC free article] [PubMed] [Google Scholar]

5. Glass JE, Hamilton AM, Powell BJ, Perron BE, Brown RT, Ilgen MA. Specialty substance use disorder services following brief alcohol intervention. Addiction. 2015;110(9):1404-1415. [PMC free article] [PubMed] [Google Scholar]

6. Institute of Medicine Broadening the Base of Treatment for Alcohol Problems: A Report of the Committee for the Study of Treatment and Rehabilitation for Alcoholism. Washington, DC: National Academies Press; 1990. [Google Scholar]

7. Institute of Medicine Improving the Quality of Health Care for Mental and Substance-Use Conditions. Washington, DC: National Academies Press; 2006. [Google Scholar]

8. Watkins K, Pincus HA, Tanielian TL, Lloyd J. Using the chronic care model to improve treatment of alcohol use disorders in primary care settings. J Stud Alcohol. 2003;64(2):209-218. [PubMed] [Google Scholar]

9. Saitz R, Larson MJ, Labelle C, Richardson J, Samet JH. The case for chronic disease management for addiction. J Addict Med. 2008;2(2):55-65. [PMC free article] [PubMed] [Google Scholar]

10. Willenbring ML. Medications to treat alcohol dependence. JAMA. 2007;298(14):1691-1692. [PubMed] [Google Scholar]

11. Knickman J, Krishnan R, Pincus H. Improving access to effective care for people with mental health and substance use disorders. JAMA. 2016;316(16):1647-1648. [PubMed] [Google Scholar]

12. Project MATCH Research Group Matching alcoholism treatments to client heterogeneity: Project MATCH three-year drinking outcomes. Alcohol Clin Exp Res. 1998;22(6):1300-1311. [PubMed] [Google Scholar]

13. UKATT Research Team Effectiveness of treatment for alcohol problems: findings of the randomised UK Alcohol Treatment Trial (UKATT). BMJ. 2005;331(7516):541. [PMC free article] [PubMed] [Google Scholar]

14. Jonas DE, Amick HR, Feltner C, et al.. Pharmacotherapy for adults with alcohol use disorders in outpatient settings. JAMA. 2014;311(18):1889-1900. [PubMed] [Google Scholar]

15. Miller WR. Sacred cows and greener pastures: reflections from 40 years in addiction research. Alcohol Treat Q. 2016;34(1):92-115. [Google Scholar]

16. National Committee for Quality Assurance HEDIS 2011 Volume 2: Technical Specifications. Washington, DC: National Committee for Quality Assurance; 2011. [Google Scholar]

17. Hepner KA, Watkins KE, Farmer CM, Rubenstein L, Pedersen ER, Pincus HA. Quality of care measures for the management of unhealthy alcohol use. J Subst Abuse Treat. 2017;76:11-17. [PMC free article] [PubMed] [Google Scholar]

18. Willenbring ML, Olson DH. A randomized trial of integrated outpatient treatment for medically ill alcoholic men. Arch Intern Med. 1999;159(16):1946-1952. [PubMed] [Google Scholar]

19. Saitz R, Cheng DM, Winter M, et al.. Chronic care management for dependence on alcohol and other drugs: the AHEAD randomized trial. JAMA. 2013;310(11):1156-1167. [PMC free article] [PubMed] [Google Scholar]

20. Watkins KE, Ober AJ, Lamp K, et al.. Collaborative care for opioid and alcohol use disorders in primary care: the SUMMIT randomized clinical trial. JAMA Intern Med. 2017;177(10):1480-1488. [PMC free article] [PubMed] [Google Scholar]

21. Bradley KA, Ludman EJ, Chavez LJ, et al.. Patient-centered primary care for adults at high risk for AUDs: the Choosing Healthier Drinking Options in Primary Care (CHOICE) trial. Addict Sci Clin Pract. 2017;12(15). [PMC free article] [PubMed] [Google Scholar]

22. West SG, Duan N, Pequegnat W, et al.. Alternatives to the randomized controlled trial. Am J Public Health. 2008;98(8):1359-1366. [PMC free article] [PubMed] [Google Scholar]

23. Bradley KA, DeBenedetti AF, Volk RJ, Williams EC, Frank D, Kivlahan DR. AUDIT-C as a brief screen for alcohol misuse in primary care. Alcohol Clin Exp Res. 2007;31(7):1208-1217. [PubMed] [Google Scholar]

24. National Institute on Alcohol Abuse and Alcoholism Helping Patients Who Drink Too Much: A Clinician’s Guide. Washington, DC: National Institutes of Health; 2005. NIH publication 07–3769. [Google Scholar]

25. Bradley KA, Kivlahan DR, Zhou XH, et al.. Using alcohol screening results and treatment history to assess the severity of at-risk drinking in Veterans Affairs primary care patients. Alcohol Clin Exp Res. 2004;28(3):448-455. [PubMed] [Google Scholar]

26. Katon WJ, Lin EH, Von Korff M, et al.. Collaborative care for patients with depression and chronic illnesses. N Engl J Med. 2010;363(27):2611-2620. [PMC free article] [PubMed] [Google Scholar]

27. Bradley KA, Kivlahan DR. Bringing patient-centered care to patients with alcohol use disorders. JAMA. 2014;311(18):1861-1862. [PMC free article] [PubMed] [Google Scholar]

28. Barry MJ, Edgman-Levitan S. Shared decision making—the pinnacle of patient-centered care. N Engl J Med. 2012;366(9):780-781. [PubMed] [Google Scholar]

30. Gastfriend DR, Garbutt JC, Pettinati HM, Forman RF. Reduction in heavy drinking as a treatment outcome in alcohol dependence. J Subst Abuse Treat. 2007;33(1):71-80. [PubMed] [Google Scholar]

31. Anton RF, O’Malley SS, Ciraulo DA, et al.; COMBINE Study Research Group . Combined pharmacotherapies and behavioral interventions for alcohol dependence: the COMBINE study: a randomized controlled trial. JAMA. 2006;295(17):2003-2017. [PubMed] [Google Scholar]

32. Cisler RA, Zweben A. Development of a composite measure for assessing alcohol treatment outcome. Alcohol Clin Exp Res. 1999;23(2):263-271. [PubMed] [Google Scholar]

33. Sobell LC, Brown J, Leo GI, Sobell MB. The reliability of the Alcohol Timeline Followback when administered by telephone and by computer. Drug Alcohol Depend. 1996;42(1):49-54. [PubMed] [Google Scholar]

34. Saha TD, Chou SP, Grant BF. Toward an alcohol use disorder continuum using item response theory: results from the National Epidemiologic Survey on Alcohol and Related Conditions. Psychol Med. 2006;36(7):931-941. [PubMed] [Google Scholar]

35. Forcehimes AA, Tonigan JS, Miller WR, Kenna GA, Baer JS. Psychometrics of the Drinker Inventory of Consequences (DrInC). Addict Behav. 2007;32(8):1699-1704. [PubMed] [Google Scholar]

36. Bertholet N, Gaume J, Faouzi M, Gmel G, Daeppen JB. Predictive value of readiness, importance, and confidence in ability to change drinking and smoking. BMC Public Health. 2012;12:708. [PMC free article] [PubMed] [Google Scholar]

37. Babor TF, Higgins-Biddle JC, Saunders JB, Monteiro MG AUDIT: the alcohol use disorders identification test: guidelines for use in primary care, 2nd edition. 2001. http://www.stir.ac.uk/social-science/. Accessed February 1, 2012.

38. DeSalvo KB, Fan VS, McDonell MB, Fihn SD. Predicting mortality and healthcare utilization with a single question. Health Serv Res. 2005;40(4):1234-1246. [PMC free article] [PubMed] [Google Scholar]

39. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 2001;16(9):606-613. [PMC free article] [PubMed] [Google Scholar]

40. Spitzer RL, Kroenke K, Williams JBW; Patient Health Questionnaire Primary Care Study Group . Validation and utility of a self-report version of PRIME-MD. JAMA. 1999;282(18):1737-1744. [PubMed] [Google Scholar]

41. Blanchard EB, Jones-Alexander J, Buckley TC, Forneris CA. Psychometric properties of the PTSD Checklist (PCL). Behav Res Ther. 1996;34(8):669-673. [PubMed] [Google Scholar]

42. Sheehan DV, Lecrubier Y, Sheehan KH, et al.. The Mini-International Neuropsychiatric Interview (M.I.N.I.). J Clin Psychiatry. 1998;59(suppl 20):22-33. [PubMed] [Google Scholar]

43. Liang JK-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13-22. [Google Scholar]

44. McCullagh P, Nelder JA. Generalized Linear Models. Vol 37 Boca Raton, FL: CRC Press; 1989. [Google Scholar]

45. White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377-399. [PubMed] [Google Scholar]

46. Little R, Rubin D. Statistical Analysis With Missing Data. 2nd ed Hoboken, NJ: John Wiley & Sons Inc; 2014. [Google Scholar]

47. Orford J, Hodgson R, Copello A, Wilton S, Slegg G, Team UR; UKATT Research Team . To what factors do clients attribute change? J Subst Abuse Treat. 2009;36(1):49-58. [PubMed] [Google Scholar]

48. Orford J, Hodgson R, Copello A, et al.; UKATT Research Team . The clients’ perspective on change during treatment for an alcohol problem. Addiction. 2006;101(1):60-68. [PubMed] [Google Scholar]

49. Moos RH. Context and mechanisms of reactivity to assessment and treatment. Addiction. 2008;103(2):249-250. [PubMed] [Google Scholar]

50. Bradley K, Williams EC. Necessary clarifications concerning results of the SUMMIT trial. JAMA Intern Med. In press. [PubMed] [Google Scholar]

51. Harris AH, Humphreys K, Bowe T, Tiet Q, Finney JW. Does meeting the HEDIS substance abuse treatment engagement criterion predict patient outcomes? J Behav Health Serv Res. 2010;37(1):25-39. [PubMed] [Google Scholar]

52. March JS, Silva SG, Compton S, Shapiro M, Califf R, Krishnan R. The case for practical clinical trials in psychiatry. Am J Psychiatry. 2005;162(5):836-846. [PubMed] [Google Scholar]

53. Wilson AD, Bravo AJ, Pearson MR, Witkiewitz K. Finding success in failure: using latent profile analysis to examine heterogeneity in psychosocial functioning among heavy drinkers following treatment. Addiction. 2016;111(12):2145-2154. [PMC free article] [PubMed] [Google Scholar]

54. Kline-Simon AH, Litten RZ, Weisner CM, Falk DE. Posttreatment low-risk drinking as a predictor of future drinking and problem outcomes among individuals with alcohol use disorders. Alcohol Clin Exp Res. 2017;41(3):653-658. [PMC free article] [PubMed] [Google Scholar]