Efficacy of a Psychosocial Pain Management Intervention for Men and Women With Substance Use Disorders and Chronic Pain: A Randomized Clinical Trial

Key Points

Question  Does an integrated behavioral pain management treatment (Improving Pain During Addiction Treatment [ImPAT]) added to standard residential addiction treatment improve pain-related and substance use–related outcomes compared with an attention-matched control condition in men and women?

Findings  In this randomized clinical trial including 510 participants, men randomized to the ImPAT condition displayed greater pain tolerance compared with men randomized to the control condition and women randomized to the ImPAT condition experienced significant improvements in pain intensity compared with women randomized to the control group over 12 months of follow-up. No differences were detected between the ImPAT and control conditions on substance use–related outcomes.

Meaning  When added to usual substance use treatment, ImPAT led to meaningful improvements in pain management for men and women, although additional services may be needed to enhance treatment outcomes in this relatively severe group of patients.

Importance  Chronic pain is common in those with substance use disorders (SUDs) and predicts poorer addiction treatment outcomes. A critical challenge for addiction treatment is to develop effective methods to improve pain-related and substance use–related outcomes for those in treatment for SUDs.

Objective  To examine the efficacy of an integrated behavioral pain management intervention (Improving Pain During Addiction Treatment [ImPAT]) for men and women with SUDs to treat pain, functioning, and substance use.

Design, Setting, and Participants  In this randomized clinical trial, 8 sessions of ImPAT were compared with 8 sessions of a supportive psychoeducational control (SPC) condition for adults with pain treated at a large residential SUD treatment program. Follow-up occurred at 3, 6, and 12 months postbaseline. A total of 1372 adults were screened, including 960 men and 412 women, and 510 adults were randomized, including 264 men and 246 women. The goal was to recruit approximately equal numbers of men and women to examine results separately in men and women. A total of 470 of 510 participants (92.2%) completed at least 1 follow-up assessment. Data were collected from October 3, 2011, to January 14, 2016. Data were analyzed from February 1, 2016, to May 1, 2020.

Interventions  ImPAT focused on how a psychosocial model of pain was associated with functioning and relapse prevention and provides skills to manage pain. SPC served as the active control condition and involved discussions of topics like nutrition and the course of addiction, which were intended to be relevant to the patient population and to have face validity but be distinct from the content of ImPAT.

Main Outcomes and Measurements  The primary outcomes were pain intensity, pain-related functioning, and behavioral pain tolerance at 12 months. Secondary outcomes were frequency of alcohol and drug use over 12 months.

Results  Of the 510 included participants, the mean (SD) age was 34.8 (10.3) years. A total of 133 men and 122 women were assigned to ImPAT, and a total of 131 men and 124 women were assigned to the SPC condition. Over 12 months of follow-up, randomization to the ImPAT intervention was associated with higher tolerance of pain among men, higher by a mean score of 0.11 (95% CI, 0.03 to 0.18; P = .004; Cohen d = 0.40) at 3 months and by 0.07 (95% CI, −0.01 to 0.19; P = .11; Cohen d = 0.25) at 12 months. Women receiving the ImPAT intervention experienced a reduction in pain intensity from 3 to 12 months, while women receiving the SPC condition experienced an increase in pain intensity, resulting in lower pain in the ImPAT condition by a mean score of 0.58 (95% CI, −0.07 to 1.22; P = .08; Cohen d = −0.22) at 12 months. No differences were found between the ImPAT and SPC conditions on alcohol or drug use.

Conclusions and Relevance  Behavioral pain management is not typically included in addiction treatment, but the present results indicate that this type of intervention was associated with better pain-related outcomes, including pain tolerance in men and pain intensity in women. Improvements in substance use–related outcomes beyond that achieved by treatment as usual were not observed. Treatment programs should consider providing psychosocial pain management services to augment standard addiction treatment.

Trial Registration  ClinicalTrials.gov Identifier: NCT01372267

As highlighted in the 2016 US Surgeon General’s Report, substance use disorders (SUDs) are common, and SUD treatment services can be effective in improving long-term outcomes.¹ However, a significant barrier is the presence of a cooccurring psychiatric or medical condition that, when poorly managed, can trigger a relapse. Chronic pain is one such condition that is common in SUD treatment settings, with an estimated prevalence of approximately 1 in 4 adults treated in residential SUD treatment settings.^2-4 Existing observational evidence indicates that individuals with persistent pain have worse drug and alcohol treatment outcomes compared with individuals without pain or with reduced pain.^5,6 These findings point to the need to improve treatments for pain among those with SUDs as a method to both improve general functioning and potentially improve posttreatment substance use outcomes.

Prescription opioids are commonly used to treat pain in the US, but their use is controversial because of the risk of misuse and association with adverse outcomes.^7-11 These concerns are greater in individuals with SUDs.^7,9 Nonpharmacological pain management approaches offer an attractive alternative. Two forms of psychotherapy, cognitive behavioral therapy (CBT) and acceptance-based treatments, have established efficacy in treating individuals with a variety of chronic pain conditions.^12-16 However, most studies exclude individuals with cooccurring SUDs, and few have examined the effect of CBT or acceptance-based approaches on substance use. The existing literature supporting psychotherapeutic pain management approaches in those with SUDs includes 2 open trials that documented improvements in pain level and substance use from pretreatment to posttreatment.^17,18 A 2019 pilot study¹⁹ of CBT in 40 methadone-maintained patients with opioid use disorders found preliminary evidence of reductions in substance use after treatment compared with a drug counseling group but no difference in pain level. Another 2019 pilot trial²⁰ in veterans supported the feasibility and initial efficacy of an integrated treatment combining acceptance and commitment therapy for pain interference with mindfulness-based relapse prevention for opioid misuse. To our knowledge, the only full-scale randomized trial examined the effect of a combined CBT and acceptance-based approach using an integrated behavioral pain management intervention (Improving Pain During Addiction Treatment [ImPAT]) in 128 veterans recruited during an episode of SUD treatment.²¹ This study demonstrated that participants in the ImPAT group had significantly less pain, improved functioning, and less alcohol use over the 12-month follow-up period than participants randomly assigned to an active control condition. However, this study had several weaknesses, including a lack of generalizability because it was conducted exclusively in veterans and had very few female participants.

The present study was undertaken to extend this work to men and women separately in a community treatment setting in a randomized clinical trial. Men and women were recruited from a large residential SUD treatment program and randomized to receive either the ImPAT condition or an attention-matched supportive psychoeducational control (SPC) condition in addition to treatment as usual. Patients were observed for a year, with pain and substance use–related outcomes evaluated at 3, 6, and 12 months.

Recruitment occurred at a large residential SUD treatment program between October 3, 2011, and September 1, 2014. The study was approved by the University of Michigan Medical School Institutional Review Board. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline, and the trial protocol can be found in Supplement 1. All eligible participants provided written informed consent before completing the screening survey. Participants were compensated up to $185 for study completion.

Participants were recruited by study staff from separate men’s and women’s units. Exclusion criteria were (1) acute suicidality, (2) cognitive impairment or psychotic symptoms, (3) inability to understand English, (4) inability to provide informed consent, and (5) age younger than 18 years. Inclusion criteria for the randomized clinical trial were (1) report of pain of at least moderate or greater intensity over the 3 months prior to baseline (score of 5 or greater on the Numeric Rating Scale of Pain Intensity²² as an average of usual and worst pain rating) and (2) a planned treatment discharge date that would allow for sufficient time for intervention delivery, as determined by completing an intake at the residential treatment program within the past 60 days allowing for at least 30 days before planned discharge.

All included participants completed a baseline assessment. Randomization conducted by a study biostatistician was stratified by sex to the ImPAT or SPC conditions during each treatment wave, such that when the ImPAT intervention was being conducted in the men’s unit, the SPC condition was being conducted in the women’s unit, and vice versa. Because the intervention was group based, group enrollment was capped at 12 participants. In instances where more than 12 participants were eligible at a given time, participants were randomly selected from the pool of all eligible screened participants for the group; when fewer than 12 participants were eligible, all participants were invited to participate in the group. Treatment center staff were blinded to participant condition.

Participants were assessed after treatment (baseline) and at 3, 6, and 12 months following enrollment. At baseline, participants completed the Working Alliance Inventory to assess perceptions of treatment and therapist bond.²³ Follow-up assessments at 3, 6, and 12 months included all main outcomes and were conducted by study staff uninvolved in participant allocation or treatment. These assessments consisted of interviews, a self-report questionnaire, a voluntary urine drug screen, and completion of a pain tolerance task, the ischemic pain task.^24,25

A sizable portion of participants (238 of 510 [46.7%]) were incarcerated at least once during the course of follow-up. We received approval from the University of Michigan Institutional Review Board and Michigan Department of Corrections to meet with incarcerated individuals, who completed only self-report survey components of the assessments.

The ImPAT and SPC conditions were designed to be identical in terms of the amount of therapist–patient contact. Both conditions involved 8 one-hour group-based sessions delivered over the course of approximately 4 weeks (eTables 1 and 2 in Supplement 2). Training for both the ImPAT and SPC conditions was similar in structure. Master’s-level therapists were trained on the specific session content and proper delivery during training workshops prior to participant recruitment. During the trial, therapists met regularly with clinical supervisors to discuss session content, process difficult cases, and review session recordings.

The ImPAT intervention used in the present study was consistent with what has been described previously,^17,21 with the exception that it was delivered over fewer sessions. Briefly, the intervention was designed to highlight the link between pain and poor functioning as well as the potential use of substances as a maladaptive coping strategy for pain. Treatment sessions explored new ways of conceptualizing and responding to pain, with the goal of also preventing relapses to substance use. Several sessions involved discussions of ways to accept or adapt to the reality of chronic pain and strategies for tolerating and accepting painful physical and emotional stimuli.²⁶

The SPC condition was similar to the ImPAT condition in terms of level of attention and group support without any of the specific elements of ImPAT that target new ways of adapting to pain or avoiding relapse to substances. The group sessions were designed around control condition content that has been used previously.^21,27,28

Pain intensity was calculated using the first item from the Numeric Rating Scale of Pain Intensity where participants provided a rating of usual pain level on an 11-point scale, where 0 indicates no pain and 10 indicates the worst pain imaginable, over the past 3 months or since the prior assessment. Pain-related functioning was measured using the general activity subsection of the West Haven-Yale Multidimensional Pain Inventory.²⁹ Items in this subsection assess the frequency in which a participant completes common everyday activities. Pain tolerance was measured using the ischemic pain task,^24,25 which involves performing hand grip exercises while blood flow to the arm is obstructed through the use of a standard blood pressure cuff. In this procedure, the nondominant arm of the participant was elevated above the heart for 30 seconds to allow for the blood to drain, after which a blood pressure cuff was positioned and inflated to approximately 200 mm Hg. Participants were then asked to perform hand grip exercises and instructed to continue until the pain in their arm become intolerable. The maximum amount of time allowed was 120 seconds.

For the secondary outcome, substance use was assessed using the Timeline Followback,^30,31 a calendar-assisted, interviewer-administered structured interview that has demonstrated strong test-retest reliability. During this interview-based assessment, participants were asked to separately recall any use of alcohol or drugs on each day during the 90 days prior to last entering a controlled environment (eg, the treatment facility) at baseline and since the prior assessment session or for the prior 90 days, whichever was shorter, for the 3-month, 6-month, and 12-month follow-up assessments. For drug use, this represents the percentage of days wherein no drugs were used across all types of drugs.

To compare the durability of treatment effects on pain-related and substance use–related outcomes separately for men and for women, the primary analysis used outcomes assessed at the posttreatment follow-up time points of 3, 6, and 12 months following enrollment as dependent variables using longitudinal data models. Prior to longitudinal modeling, all measures were evaluated for their distributional characteristics. All analyses of outcomes used an intent-to-treat framework and were done separately by sex. Linear mixed-effects models, with an autoregressive covariance structure for within-person time effects and an unstructured covariance structure for random intercepts for group, were used to assess the durability of the outcome measures. Intraclass correlations for group by condition ranged from 0.001 to 0.11. Condition, time, and the time by condition interaction were evaluated for significance in all models. (In addition to the primary models, an additional set of analyses were conducted that also adjusted for baseline values of the outcome measure. The results of these analyses were essentially the same and are available from the authors on request.) Time was coded as 0, 3, and 9 for 3-month, 6-month, and 12-month follow-up times, respectively. F statistics were used in all mixed models. χ² and t tests were used to compare discrete and continuous measures, respectively, in cross-sectional comparisons. Statistical significance was considered at an α of .005 to account for multiple comparisons, and all P values were 2-tailed. All analyses were conducted using SAS version 9.4 (SAS Institute).

For the screening portion of the study, data were collected from 1396 participants, of which 767 were eligible to participate. We initially enrolled 512 participants into the randomized clinical trial. Two enrolled participants were excluded from the study prior to randomization, including 1 due to leaving the treatment site before the start of the trial and 1 due to previously unreported cognitive impairment. Of the 510 included patients, the mean (SD) age was 34.8 (10.3) years, and 264 were men and 246 were women. A total of 133 men and 122 women were assigned to ImPAT, and 131 men and 124 women were assigned to the SPC condition (Figure 1). Overall, 470 of 510 participants (92.2%) completed at least 1 of the follow-up assessments; 395 participants (80.0%), including 214 men and 181 women, completed the 3-month follow-up; 424 participants (87.7%), including 226 men and 198 women, completed the 6-month follow-up; and 429 participants (91.7%), including 228 men and 201 women, completed the 12-month follow-up assessment.

Baseline characteristics of the male and female samples are shown in Table 1. As expected in a residential substance use treatment facility, the sample had extensive and complex substance use presentations, with 219 of 264 men (83.0%) and 222 of 246 women (90.2%) likely meeting criteria for more than 1 substance disorder based on the Alcohol, Smoking and Substance Involvement Screening Test³² (score greater than 10 for alcohol; score greater than 3 for other substances). The 3 most commonly used substances were cannabis, opioids, and alcohol for men and opioids, cocaine, and heroin for women. Back pain was the most frequently reported site of pain in both men (201 [76.1%]) and women (209 [85.0%]). Men also frequently reported knee pain (134 [50.8%]), leg or foot pain (124 [47.0%]), and hand pain (102 [38.6%]), and women frequently reported leg or foot pain (136 [55.3%]), knee pain (130 [52.8%]), and head, face, or jaw pain (130 [52.8%]).

Overall, 223 of 264 men (84.5%) and 209 of 246 women (85.0%) attended 6 or more of the 8 sessions. For the ImPAT condition, 94 of 133 men (70.7%) and 88 of 122 women (72.1%) attended all 8 sessions, with only 2 men (1.5%) and 5 women (4.1%) attending no intervention sessions. In the SPC condition, 75 of 131 men (57.3%) and 83 of 124 women (66.9%) attended all 8 sessions, with 4 men (3.1%) and 2 women (1.6%) attending no sessions. Men attended a mean (SD) of 7.23 (1.66) sessions and 6.64 (2.22) sessions in the ImPAT and SPC conditions, respectively. Women attended a mean (SD) of 7.15 (2.00) sessions and 6.84 (2.09) sessions in the ImPAT and SPC conditions, respectively. Participants in both conditions reported similar perceptions of treatment and therapist bond; the mean (SD) Working Alliance Inventory scores for ImPAT were 56.53 (11.17) and 56.51 (11.81) for men and women, respectively, and for SPC were 54.60 (11.12) and 56.38 (11.57) for men and women, respectively.

Table 2 reports results of longitudinal modeling of outcome measures and effect sizes. A significant time by condition interaction was found in women but not in men for pain intensity as measured by the first item from the Numeric Rating Scale of Pain Intensity; in women, a greater reduction in pain intensity was seen in the ImPAT condition compared with the SPC condition over the course of the follow-up period (β, −0.11; SE, 0.03; 95% CI, −0.17 to 0.05; P = .001), resulting in lower pain in the ImPAT compared with the SPC condition by 0.58 (95% CI, −0.07 to 1.22; P = .08; Cohen d = −0.22) at 12 months (Figure 2). Men showed significantly greater pain tolerance as measured by the ischemic pain task in the ImPAT condition compared with the SPC condition over the course of follow-up (β, 0.11; SE, 0.04; 95% CI, 0.03 to 0.16; P = .004), higher by 0.11 (95% CI, 0.03 to 0.18; P = .004; Cohen d = 0.40) at 3 months and by 0.07 (95% CI, −0.01 to 0.19; P = .11; Cohen d = 0.25) at 12 months.

No significant effects of condition were detected for alcohol or drug outcomes across the follow-up assessments (Table 2). Figure 3 illustrates the patterns over time with observed mean values at baseline and model-based means at the follow-up times. See eTable 3 in Supplement 2 for effect sizes between conditions by outcome and time and the eAppendix in Supplement 2 for additional analyses looking at effects of incarceration and missing data on outcomes.

Chronic pain is common among those with SUDs,^2-4 and addiction treatment professionals often struggle with how to manage pain in their patients.³³ The present study recruited adults with pain receiving residential treatment for SUDs and examined the efficacy of a psychological integrative pain management and relapse prevention approach (ImPAT) compared with an attention-matched control condition (SPC) on pain-related and substance use–related outcomes over a 12-month period. Compared with those receiving the SPC condition, men receiving the ImPAT intervention displayed greater behavioral pain tolerance over follow-up, and women receiving the ImPAT intervention reported significantly greater reductions in pain intensity over the latter part of follow-up. No effects were seen on days of alcohol or drug use during the follow-up period. These results are largely consistent with, to our knowledge, the only other published trial powered to test efficacy of a psychosocial pain management approach in SUD treatment.²¹ Notable findings are the potential of these interventions to improve pain intensity and tolerance, the different effects observed in women and men, and the lack of effects detected on substance use outcomes.

CBT and acceptance-based treatments for pain management are empirically supported^12-16; however, prior trials of these interventions have typically excluded those with SUDs and/or have not examined SUD-related outcomes. Pilot studies have highlighted the potential viability of psychosocial approaches for individuals being treated for SUDs.^19,34 One prior study found evidence of the efficacy of this psychotherapeutic pain management approach in veterans receiving SUD treatment.²¹ The present study involves a strong test of ImPAT by comparing it with a contact-matched, psychoeducational control condition, with both conditions overlaid onto an episode of residential SUD treatment. A strength of this study is the recruitment of approximately equal numbers of men and women to examine sex-specific effects separately.

This study included a behavioral measure of pain tolerance. Low pain tolerance has received attention in the methadone maintenance literature^35,36 but is relatively understudied in broader SUD literature. The present study tracked this outcome and found that pain tolerance was improved in men receiving ImPAT compared with the control condition. Prior work has documented that men have higher pain tolerance than women^37,38 and that individuals with SUDs are less tolerant of pain than those without SUDs.³⁹ Few nonpharmacological treatments are available to improve pain tolerance,⁴⁰ and to our knowledge, none have been identified for those with SUDs. Future work should incorporate behavioral measures of pain tolerance to elucidate the role of this construct in sustained recovery.

Women receiving the ImPAT intervention reported greater improvements in pain intensity during the latter part of follow-up than those receiving the control condition. On average, women in the ImPAT condition reported 0.58 (95% CI, −0.07 to 1.22) lower pain intensity as measured by the first item from the Numeric Rating Scale of Pain Intensity on a scale from 0 to 10 at the 12-month follow-up compared with those receiving the SPC condition, a difference that is consistent with the small effect size but nonetheless noteworthy in this highly impaired study population. Given the consistent connection between pain level and numerous indicators of quality of life and mental health,^41,42 these longer-term improvements are likely to have a meaningful impact on the lives of these women.

No significant differences were found between conditions in men or women on percentage of days abstinent from alcohol or drug use. This differs from a prior trial of the ImPAT intervention in veterans,²¹ which found benefits of this intervention approach on percentage days abstinent from alcohol. The setting for the present study is based in a more intensive addiction treatment setting than the prior study (residential vs intensive outpatient treatment), and the participants likely had greater severity of SUDs, poorer supports for sobriety, and greater criminal justice involvement than those in the prior study. This is reflected in the high number of participants who were incarcerated during the follow-up period in the present study. These factors could have made it difficult to observe the effect of an adjunctive treatment service, such as ImPAT, on substance use–related outcomes. In addition, a good deal of variability was seen in self-report of days of use with regard to different substances at baseline, and the measurement of drug use likely lacked precision in the variable summarizing use of all substances. Opioid use disorders in particular may be especially important given the exceptionally high rate of chronic pain in this subpopulation. One approach is to prevent progression toward opioid use disorders among those at particularly high risk. A 2019 study⁴³ of patients prescribed opioids for chronic pain used a mindfulness-oriented therapy and mitigated opioid misuse risk in addition to reducing pain and improving psychological health. Another approach focuses on treating opioid use disorders when they occur. Two recent pilot studies have demonstrated initial acceptability, feasibility, and preliminary efficacy among individuals using opioids; one treatment provided CBT for pain management in methadone-maintained patients³⁴ and the other provided integrated acceptance and commitment therapy and mindfulness-based relapse prevention to veterans with chronic pain and hazardous opioid use.²⁰ These studies, along with the current findings, highlight promising psychosocial treatments to improve outcomes in those with chronic pain and opioid misuse. Of note, medication-assisted therapy has strong support for the treatment of opioid use disorders,⁴⁴ and the present study was based in an abstinence-based residential program. Future trials could clarify whether effects on pain-related variables (eg, intensity, functioning, tolerance) are more pronounced when psychosocial treatments are added to medication-assisted therapy for opioid use disorders.

Several limitations of this study warrant noting. Few details are known about the causes or nature of participants’ pain at baseline or the timing of the pain condition relative to the development of SUDs, and these factors may be important determinants of intervention efficacy. Furthermore, although study retention was strong, with more than 90% of participants completing at least 1 of the follow-up assessments, the possibility of differential treatment effects of those lost to follow-up may not be captured in the current findings. Many of the measures are based on self-report; some individuals may have overreported or underreported pain or substance use, potentially limiting the accuracy of measurement. Statistically significant outcomes (ie, pain intensity in women and pain tolerance in men) have modest effect sizes. In addition, the original analysis plan was modified to fit the distribution of the data among other factors. The protocol was used at a single residential SUD treatment program. The effect of the theoretical orientation of the treatment program on treatment effects reported in this study is unclear. In addition, participants had very severe SUDs and extensive engagement in the criminal justice system, where controlled access to substances and limited opportunities to use behavioral tools for pain management may affect findings. This likely made it difficult to detect intervention effects on substance use, and effects may be larger on pain-related outcomes in other settings.

These findings highlight the potential benefits of this approach on pain-related outcomes for SUD treatment settings and raise the possibility that integrating behavioral pain management and SUD services could be particularly beneficial for the large number of individuals with cooccurring pain and SUDs. However, the effect sizes were small, and we did not detect an effect of the ImPAT condition on SUD outcomes, highlighting the need for future work to continue to look at ways to enhance integrated pain and SUD treatment effects to improve clinical outcomes in the high-risk group of patients with both pain and SUDs.

Accepted for Publication: June 9, 2020.

Corresponding Author: Mark A. Ilgen, PhD, VA Center for Clinical Management Research (CCMR), Department of Veterans Affairs Healthcare System, 2800 Plymouth Rd, Ann Arbor, MI 48109 (marki@med.umich.edu).

Published Online: July 29, 2020. doi:10.1001/jamapsychiatry.2020.2369

Author Contributions: Ms Jannausch had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Ilgen, Chermack, Kim, Blow.

Acquisition, analysis, or interpretation of data: Ilgen, Coughlin, Bohnert, Chermack, Price, Kim, Jannausch.

Drafting of the manuscript: Ilgen, Coughlin, Price, Jannausch.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Ilgen, Coughlin, Kim, Jannausch.

Obtained funding: Ilgen, Chermack, Kim, Blow.

Administrative, technical, or material support: Ilgen, Price.

Study supervision: Ilgen, Price.

Conflict of Interest Disclosures: Dr Ilgen is a part of ArborSense, which is developing a wearable alcohol sensor, and consults for Northrop Grumman on a project to prevent suicide in military personnel. Dr Blow has received personal fees from the Hazelden Betty Ford Foundation. No other disclosures were reported.

Funding/Support: This work was supported by grant R01 DA029587 from the National Institute on Drug Abuse. Dr Coughlin’s time was supported by grant T32AA007477 from the National Institute on Alcohol Abuse and Alcoholism.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this report are those of the authors and do not necessarily represent those of the National Institute on Drug Abuse or the National Institute on Alcohol Abuse and Alcoholism.

Data Sharing Statement: See Supplement 3.

Additional Contributions: We thank the patients and staff at Community Programs Inc for their support of this project.

Additional Information: Dr Chermack died in 2019.