A Randomized Clinical Trial of Group versus Standard Behavioral Couples Therapy Plus Individually Based Treatment for Patients with Alcohol Dependence

Abstract

Objective

Multiple studies show that behavioral couples therapy (BCT) is more efficacious than individually-based therapy (IBT) for substance use and relationship outcomes among patients with alcohol use disorder. To facilitate dissemination, a multi-couple, rolling admission Group BCT (G-BCT) format has been suggested as an alternative to the one couple at a time, conjoint Standard BCT (S-BCT) format. This randomized study compared outcomes of G-BCT versus S-BCT over a 1-year follow-up. We predicted that G-BCT, as compared to S-BCT, would have equivalent (i.e., non-inferior) improvements on substance and relationship outcomes.

Method

Participants were patients (N = 101) with alcohol dependence and their heterosexual relationship partners without substance use disorder. Participants were mostly White, in their forties, and 30% of patients were women. Patients were randomized to either G-BCT plus 12-step-oriented IBT or S-BCT plus IBT. Primary outcomes included: Timeline Followback Interview percentage days abstinent and Inventory of Drug Use Consequences measure of substance-related problems. Secondary outcome was Dyadic Adjustment Scale. Outcome data were collected at baseline, post-treatment, and quarterly for 1-yr follow-up.

Results

Results overall found no support for the predicted statistical equivalency of G-BCT and S-BCT. Rather than the predicted equivalent outcomes, substance and relationship outcomes were significantly worse for G-BCT than S-BCT in the last 6–9 months of the 12-month follow-up period, because G-BCT deteriorated and S-BCT maintained gains during follow-up.

Conclusion

This was the first study of the newer rolling admission group format for BCT. It proved to have worse not equivalent outcomes compared to standard conjoint BCT.

Behavioral couples therapy (BCT) is a couple-based therapy for adults with alcohol use disorder (AUD) that aims to help the AUD patient and their partner build support for the patient’s abstinence and improve relationship functioning. Research supports the efficacy of BCT. BCT produces greater abstinence, fewer substance-related problems, and better relationship functioning than more typical individual-based treatment for married or cohabiting AUD patients (O’Farrell & Clements, 2012). Two meta-analyses showed a medium effect size favoring BCT over individual treatment on these outcomes (Meis et al., 2013; Powers, Vedel & Emmelkamp, 2008), and recent studies have extended BCT efficacy to women AUD patients (e.g., McCrady, Epstein, Cook, Jensen, & Hildebrant, 2009; Schumm, O’Farrell, Kahler, Murphy, & Muchowski, 2014).

Government and professional guidelines have recognized BCT research support. BCT for AUD was designated “an established treatment with strong research support” by the American Psychological Association’s Division 50. Further, BCT was listed in SAMHSA’s National Registry of Evidence-Based Programs. Finally, BCT was recommended for use in AUD treatment provided by the U.K.’s National Health Service and the U.S. Dept. of Veterans Affairs.

Despite its demonstrated efficacy, BCT is not well-utilized. In their survey of addiction treatment programs, McGovern and colleagues (McGovern, Fox, Xie & Drake, 2004) found that BCT was one of the least used of the evidence based practices surveyed. A more recent national survey of VA outpatient substance use disorder (SUD) programs (Gifford, 2012) found that only 6 programs were using BCT.

Studies have identified barriers to utilization of BCT from the treatment program perspective. The cost of delivering BCT is a barrier to adopting BCT (Gifford, 2012; Schonbrun et al., 2012). Most treatment programs have resources to provide a limited amount of treatment sessions, so that funding devoted to BCT sessions must be taken from resources otherwise devoted to individual or group counseling for the patient alone. Further, the format of BCT presents challenges for many treatment programs. For example, most substance abuse programs use primarily group counseling whereas BCT is generally provided to one couple at a time (Schonbrun et al., 2012).

One strategy to overcome these barriers is to modify BCT to reduce its cost and change its format to make it more appealing and a better fit for AUD treatment programs. To address these issues, we investigated a couples group format to deliver BCT. In earlier work, we developed a 10-week BCT group with a number of conjoint sessions added before to prepare couples and afterwards to prevent relapse (e.g., O’Farrell, 1993a, 1993b). While the outcomes were favorable (O’Farrell, Choquette & Cutter, 1998), the added conjoint sessions and the closed group format reduced its appeal for community agencies that were looking for briefer treatments and generally running ongoing groups with new members added regularly.

To overcome these problems, we developed a 10-session ongoing BCT group format that has rotating content and rolling admissions in which couples join the group, complete 10 sessions, and “graduate”. A major advantage of this ongoing group format for BCT is that it fits with how other types of groups generally are run in substance abuse programs. Delivering BCT in a couples group format, rather than the one couple at a time standard conjoint BCT format, addresses both the format and cost barriers to BCT adoption noted earlier. Group BCT fits with the widespread use of group counseling in substance abuse programs and follows requests by clinicians surveyed (Gifford, 2012). Finally, an important part of the scientific rationale for evaluating group BCT is that the efficacy of group therapy has been established for the treatment of AUDs, without the presence of an intimate partner. For example, Bowen et al. (2014) found that cognitive-behavioral group therapy that did not incorporate partners into the treatment was an efficacious treatment for substance use disorders. In addition, a meta-analysis of 26 studies of relapse prevention for substance use disorders found that group and individually-based, cognitive-behavioral therapies produced similar significant improvements in substance use outcomes (Irvin, Bowers, Dunn, & Wang, 1999).

The present study is the first study of a rolling admission Group BCT (G-BCT) for AUD patients. It compares G-BCT with standard conjoint BCT (S-BCT). It is important to determine whether G-BCT, an innovation being proposed to facilitate dissemination of BCT, is equally effective as S-BCT which it is being proposed to replace. A G-BCT format, that retained the effectiveness of standard BCT and could be more easily integrated into existing services, would more likely be attractive to community substance abuse programs and thus may be more deployable in these settings.

The present study tested group couples therapy as a method to overcome cost and format barriers to BCT from the treatment program perspective. Therefore, it was important in the present study, that treatment conditions resembled how BCT would be used in many SUD treatment programs. In such programs, BCT frequently is delivered in combination with an individual-based treatment (IBT) in the form of group or individual counseling sessions for the patient. We chose a 12-step oriented group counseling IBT intervention because it reflects the dominant IBT in the U.S. (Roman & Johnson, 2004). However, it should be noted that those studying BCT outside of a SUD treatment program (e.g. in a research clinic where patients were drawn mainly from advertisements), have used a stand alone BCT model without other treatment (e.g., Epstein et al., 2007; McCrady et al., 2009; Walitzer, Dermen, Shyalla, & Kubiak, 2013).

In the present study, married or cohabiting patients with AUD were recruited from a SUD treatment program. They were randomly assigned to either (a) multi-couple, rolling admission Group BCT (G-BCT) plus 12-step oriented group IBT or (b) one couple at a time, conjoint Standard BCT (S-BCT) plus 12-step oriented group IBT. Outcome data were collected at baseline, post-treatment, and at 3-, 6-, 9- and 12-month follow-up. This study compared outcomes of G-BCT versus S-BCT over a 1-year follow-up period. We predicted that AUD patients who received G-BCT plus IBT, as compared to those who got S-BCT plus IBT, would have equivalent (i.e., non-inferior) improvements on primary outcomes of days abstinent from alcohol and drugs and substance-related problems. We also predicted that G-BCT plus IBT would have equivalent improvements as S-BCT plus IBT on the secondary outcome of relationship satisfaction. Finally, we examined whether both S-BCT and G-BCT patients significantly improved from before to after treatment and whether this improvement was maintained during the 1-year follow-up period.

Method

Institutional review boards at Harvard Medical School and at VA Boston approved this study.

Participants

Participants were heterosexual couples consisting of 101 patients with alcohol dependence and their relationship partners. They were recruited from patients seeking treatment at a large SUD treatment center in the northeastern U.S. from January 2011 to July 2013. Eligibility criteria were as follows: (1) both patient and partner were 18 years of age or older; (2) patients met past 12 month alcohol dependence diagnosis according to the Structured Clinical Interview for the DSM-IV (SCID; First, Spitzer, Gibbons & Williams, 1996) and comorbid drug use disorders also were permitted; (3) patients consumed alcohol in the 60 days prior to the study; (4) patient’s primary drug of abuse was alcohol according to an algorithm described in Fals-Stewart (1996); (5) patients did not exhibit current alcohol or drug dependence that required inpatient treatment or medical detoxification, with the understanding that after completing needed detoxification or inpatient treatment they may be eligible for the study; (6) during study-based treatment, patients were agreeable to the goal of abstinence, and willing to forgo other professional alcoholism counseling other than treatment required for a clinical emergency or to address clinical deterioration or self-help meeting attendance; (7) other than nicotine dependence, partners did not meet past 12 month diagnosis for a substance use disorder according to the SCID (First et al., 1996); (8) neither patient or partner met criteria for psychotic disorder according to the SCID (First et al., 1996); (9) neither patient or partner had a history of drug overdose or suicide attempt in the past 30 days, or were at immediate risk for homicide or suicide; (10) couple married for at least 1 year or living together in a stable common-law relationship for at least 2 years; (11) couple lived apart for no more than 4 out of the past 12 months; (12) couple had no immediate plans to separate or divorce; (13) on brief intimate partner violence (IPV) questions in the study screening interview, patient and partner denied severe IPV (i.e., that which had resulted in injury) in the past 3 years, and neither patient or partner reported fear that couples counseling might create an undue risk of violence.

Participants of which 29.7% were female patients were, on average, in their forties (patients M = 48.34, SD = 9.19; partners M = 47.68, SD = 9.96) with about 15 years of education (patients M = 14.73, SD = 2.74; partners M = 15.16, SD = 3.10). Most couples (87.1%) were married and had been married or cohabitating for around 18 years (M = 17.96, SD = 10.94). Participants’ ethnicity was mostly White (patients = 98.0%, partners = 97.0%). Patients’ annual gross median income was $44,000. Over 60% of the patients were employed full- (50.5%) or part-time (11.9%). Patients reported an average of 14.81 years of problematic alcohol use (SD = 11.10), and all had a DSM-IV diagnosis of current alcohol dependence. A minority of patients had a past 12 month drug use disorder (5.0%). Patients’ scores on the Inventory of Drug Use Consequences - Lifetime (M = 34.66, SD = 5.42) were similar to those previously found among patients entering outpatient substance use treatment (e.g., Tonigan & Miller, 2002). Finally, participants in G-BCT did not differ from those in S-BCT on the variables just described (see Table 1).

Table 1.

Pretreatment Characteristics for Participants by Treatment Condition

Characteristic	S-BCT (n = 51) M(SD) or No. (%)	G-BCT (n = 50) M(SD or No. (%)	t or χ2	p
Patients with female gender	15 (29)	15 (30)	0.004	0.948
Patient age	48.9 (8.8)	47.8 (9.6)	0.58	0.564
Patient education	14.3 (2.7)	15.2 (2.7)	1.64	0.104
Partner age	48.8 (10.1)	46.6 (9.8)	1.12	0.264
Partner education	14.9 (3.4)	15.4 (2.7)	0.84	0.404
Years married or cohabiting	18.0 (11.1)	17.9 (10.9)	0.05	0.957
Patient income (thousands/yr)- median	42.5	45.0	---a	1.00
Patient ethnicity:			---b	0.243
White	51 (100)	48 (96)
African-American	0	0
Hispanic	0	0
Asian	0	1 (2)
Other	0	1 (2)
Partner ethnicity:			---b	0.368
White	50 (98)	48 (96)
African-American	0	0
Hispanic	1 (2)	0
Asian	0	0
Other	0	2 (4)
Patient employment:			---b	0.146
Not employed	16 (31)	9 (18)
Employed full-time	27 (53)	24 (47)
Employed part-time	4 (8)	8 (16)
Other	4 (8)	9 (18)
Patient with alcohol dependence dx	51 (100)	50 (100)	---	---
Years patient problematic alcohol use	14.9 (11.6)	14.7 (10.7)	0.08	0.935
Patient other substance use dx:
Sedative/Hypnotic/Anxiolytics	0	0	---	---
Cannabis	1 (2)	2 (4)	---b	0.617
Stimulants	0	0	---	---
Opiates	0	3 (6)	---b	0.118
Cocaine	0	0	---	---
Hallucinogens	0	0	---	---
Patient InDUC-Lifetime total score	34.7 (5.6)	34.7 (5.3)	0.003	0.997

Note. S-BCT = standard behavioral couples therapy. G-BCT = group behavioral couples therapy.

^aThe p-value for this variable comes from the Independent Samples Median Test.

^bThe p-value for this variable comes from Fisher’s Exact Test because the data did not meet assumptions for chi-square.

Measures

Unless otherwise specified, measures were administered to both the patient and his/her partner at pre- and post-treatment and then again at 3-, 6-, 9-, and 12-months following the scheduled end of treatment. Each participant was paid $40 dollars for each assessment and study treatment was free.

Timeline Followback Interview (TLFB; Sobell & Sobell, 1996)

The TLFB uses a calendar and other memory aids to gather retrospective information about substance use behaviors over a specified period of time. The TLFB, which is widely used in alcoholism treatment research, has shown test-retest reliability of ≥ .80 for alcohol and illicit drugs (Sobell & Sobell, 1996). Both patient and partner completed the TLFB with reference to the patient’s behavior. Percentage days abstinent (PDA) was calculated by dividing the number of days on which the patient was not in a hospital or jail for alcohol-related reasons and they remained abstinent from alcohol and other drugs by the total days during a given time period. There were large correlations between patient- and partner-reported PDA at all time periods (r = .74–.88). To reduce possible underreporting of the patients’ substance use, we used the lower reported PDA when both partners’ data were available. When data was available from only one partner, we used the available partner’s report.

Inventory of Drug Use Consquences (InDUC; Tonigan & Miler, 2002)

The InDUC is a 45-item self-report measure of adverse consequences of alcohol and drug use. The InDUC is shown to exhibit excellent test-retest reliability, acceptable convergent validity, and good sensitivity to change in response to treatment (Tonigan & Miller, 2002). Internal reliabilities in the present study were excellent at all time periods (α > .92). Both patient and partner responded to the InDUC with reference to consequences of the patient’s alcohol and drug use. At baseline, the current (past 3 months) and lifetime versions of the InDUC were both administered. At post-treatment, the InDUC referenced the time during treatment, and the InDUC referenced the prior 3 months during the other follow-up assessments. To reduce possible underreporting of the patients’ substance-related problems, we used the higher report when both partners provided responses to an InDUC item. When data was available from only one partner on an InDUC item, we used the available partner’s report. For the lifetime version of the InDUC, response options are 0 = no or 1= yes, and possible scores range from 0–45 (see Table 1). For the current version of the InDUC, response options range from 0 = never to 3 = daily or almost daily, and possible scores range from 0–135 (see Table 2).

Table 2.

Substance Use Outcomes Observed M (SD), Sample Size, and Effect Size Change from Baseline by Treatment Condition

	Percentage days abstinent (PDA)
Treatment	Baseline M (SD)	Post M (SD)	Post d	3-month M (SD)	3-month d	6-month M (SD)	6-month d	9-month M (SD)	9-month d	12-month M (SD)	12-month d
S-BCT	30.83 (32.81) n = 51	84.67a (25.82) n = 51	1.49	78.32a (35.61) n = 51	1.12	80.87a (29.78) n = 50	1.27	84.21a (27.74) n = 49	1.25	82.51a (30.85) n = 49	1.38
G-BCT	20.58 (21.30) n = 50	84.65a (30.46) n = 50	1.62	80.06a (30.15) n = 47	1.55	71.02a (37.50) n = 45	1.05	70.74a (35.17) n = 45	1.05	69.64a (38.99) n = 45	0.98

	Inventory of Drug Use Consequences (InDUC)
Treatment	Baseline M (SD)	Post M (SD)	Post d	3-month M (SD)	3-month d	6-month M (SD)	6-month d	9-month M (SD)	9-month d	12-month M (SD)	12-month d
S-BCT	64.73 (21.59) n = 51	26.69a (30.13) n = 51	1.18	22.36a (28.16) n = 50	1.43	20.80a (30.91) n = 50	1.33	17.50a,b (21.62) n = 50	1.52	15.26a,b (22.86) n = 49	1.68
G-BCT	63.23 (19.95) n = 50	22.19a (27.39) n = 48	1.57	28.36a (28.79) n = 45	1.18	30.40a (31.50) n = 44	1.13	30.23a (32.07) n = 44	1.04	31.86a (33.95) n = 44	0.91

Note. S-BCT = standard behavioral couples therapy. G-BCT = group behavioral couples therapy. Post = post-treatment.

^aThis score shows a significant (p < .01) improvement vs. baseline.

^bThis score shows a significantly more favorable outcome for S-BCT than G-BCT at this time period. Cohen’s d effect sizes guidelines are small = 0.2, medium = 0.5, large = 0.8.

Dyadic Adjustment Scale (DAS; Spanier, 2001)

The DAS is a widely used 32-item self-report measure of overall relationship adjustment that is designed to assess relationship consensus, cohesion, satisfaction, and affectional expression. Possible total scores range from 0 to 146. The DAS exhibits excellent test-retest reliability as well as strong concurrent and criterion-related validity in differentiating distressed from non-distressed couples (Spanier, 2001). In the present study, internal reliabilities were excellent at all time periods (α > .93).

Client Satisfaction Questionnaire-8 (CSQ-8; Attkinsson & Greenfield, 2004)

The CSQ-8 is an 8 item measure that was used to assess satisfaction with study-based treatment. The CSQ-8 was administered to the patient and the partner at post-treatment. Possible total scores range from 8 to 32. The CSQ-8 has been shown to perform consistently across a range of treatment settings and has demonstrated construct validity (Attkinsson & Greenfield, 2004). In the present study, the internal reliability was excellent (α = .92).

Non-study-based treatment

Patients were interviewed about substance-related treatment that they received during the 60 days prior to joining the study. Non-study-based treatment was defined as the total number of days of hospitalization for detoxification, residential substance-related treatment, and day treatment/intensive outpatient treatment.

Procedure

Married and cohabiting male and female patients seeking treatment for an AUD (N = 964) completed a screening questionnaire to determine possible study eligibility. Based on screening questionnaire responses, 296 patients were ineligible and 668 were potentially eligible. Research staff tried to contact potentially eligible patients but were unable to reach 109 of them. Of the 559 potentially eligible patients who spoke with staff, 361 reported they were not interested in the study and were not further assessed. ¹ Another 65 reported initial interest in the study, but staff were unable to reconnect with them to confirm final interest and eligibility. For those who were potentially eligible, reported interest in the study, and responded to staff, study screening interviews were privately and separately completed with the patient and his/her partner to further assess eligibility. Of the couples who completed these interviews, 8 couples were determined to be ineligible. Participant flow into the trial is shown in Figure 1.

Figure 1.

CONSORT (Consolidated Standards of Reporting Trials) flowchart. G-BCT = group behavioral couples therapy. S-BCT = standard BCT. Post = post-treatment.

A total of 125 couples met study criteria, signed informed consent, and were scheduled to complete baseline assessments with a research assistant. Following baseline assessments, the couple’s first treatment session was scheduled. Randomization occurred after baseline assessments were completed and prior to any treatment attendance. Of the 125 consented couples, 23 couples were not randomized because they dropped out prior to completing their baseline assessment (n = 13) or prior to attending their first treatment session (n = 10). The remaining 102 couples were put into an urn randomization computer program (Stout, Wirtz, Carbonari, & Del Boca, 1994) designed to balance the treatment groups in terms of patient gender, age (34 or younger versus 35 or older), and married versus cohabitation status. Treatment assignment was concealed from the couple until they arrived for their first treatment session. One couple was removed from the study by the principal investigator after randomization and treatment initiation due to the patient’s daily very heavy drinking (e.g., BAC > 0.30) and unwillingness to undertake a higher level of care. Given these factors, the PI determined that the couple could not be treated safely under the outpatient study protocol. The patient and partner were not followed further. All of the remaining couples who attended one or more treatment sessions became part of the intent-to-treat sample; they were randomized, followed and included in the analyses as long as at least one member of the couple provided some follow-up data (N =101; see Figure 1).

Treatment Conditions

Patients in the two treatment condition were assigned to receive a total of 23 therapy sessions over the course of 12 weeks. All patients were intended to participate in 12 weekly 12-step oriented group IBT sessions. In addition, patients were intended to participate in the condition-specific treatment (i.e., multi-couple Group BCT or Standard BCT) for 11 consecutive weeks during the course of the 12 week period. Figure 2 and the text below provide more details.

Figure 2.

Experimental Design and Treatment Conditions (all patients were scheduled to receive 23 counseling sessions over a 12-week period 12 GDC sessions plus 11 condition specific sessions)

Note: GDC = Group Drug Counseling. Intro = introductory session with the patient to introduce them to joining a GDC group or with the couple to introduce them to starting G-BCT or S-BCT.

Standard Behavioral Couples Therapy (S-BCT) plus IBT Condition

For the 23 sessions conducted as part of the S-BCT condition, both patient and partner were intended to participate in 11 60-minute conjoint S-BCT sessions, which consisted of 2 introductory couple sessions to negotiate a BCT Recovery Contract and 9 more BCT couple sessions. In the remaining 12 sessions, the patient was intended to participate in 12-step oriented group IBT sessions.

S-BCT sessions were drawn from the O’Farrell and Fals-Stewart (2006) 12-session BCT manual. The 2 introductory S-BCT sessions negotiated a BCT Recovery Contract which included a daily “trust discussion” in which the patient stated an intent to stay abstinent that day and the spouse expressed support for the patient’s efforts. The other 9 S-BCT sessions included material to increase positive feelings, shared activities, and constructive communication.

Group Behavioral Couples Therapy (G-BCT) plus IBT Condition

For the 23 sessions conducted as part of the G-BCT condition, both patient and partner were planned to attend 11 couple sessions, which consisted of 2 60-minute conjoint introductory sessions to negotiate a BCT Recovery Contract and 9 90-minute multi-couple group BCT sessions together with 3 to 5 other couples. G-BCT group sessions had semi-rolling admission; couples could be placed in the group every third session, and they completed after a sequence of 9 groups total. In the remaining 12 sessions, the patient was intended to participate in 12-step oriented group IBT as described below.

G-BCT sessions were drawn from the O’Farrell and Fals-Stewart (2006) 12-session BCT manual. G-BCT used the same content as S-BCT, but adapted it for use in a manualized multi-couple group format (Fals-Stewart, O’Farrell, Golden & Birchler, 2004). Both G-BCT and S-BCT included the BCT Recovery Contract to reinforce sobriety and modules to increase positive feelings, shared activities, and constructive communication. G-BCT, as compared to S-BCT, placed greater emphasis on modeling BCT skills, with group members practicing skills with feedback from members and the group leader.

Twelve-Step Oriented Group IBT Sessions: Treatment-As-Usual (TAU) for Both Conditions

These sessions were drawn from the group drug counseling (GDC) manual (Daley, Mercer & Carpenter, 2002), slightly modified to focus on alcohol dependence; as noted in the manual, such modification is acceptable. The 12 GDC sessions were received by patients in both treatment conditions, and partners did not participate in any of the patients’ GDC sessions. Patients had one 60-minute individual introductory session, followed by 11 90-minute group sessions. This was a semi-rolling admissions group, with patients permitted to enter every 3^rd to 4^th group session (to permit some flexibility) and completing after a sequence of 11 groups. Thus, there were a total of 12 GDC sessions (11 groups, 1 introductory individual session).

GDC is based on the concept that alcoholism is a spiritual and medical disease, consistent with the philosophy of Alcoholics Anonymous (AA), and that recovery is a gradual process achieved by staying sober and attending AA self-help groups. GDC had better outcomes than professional psychotherapy in NIDA’s Collaborative Cocaine Treatment Study (Crits-Christoph, et al., 1999). The National Program Survey of Treatment Centers (Roman & Johnson, 2004) indicated that over 80% of U.S. programs ascribe to the disease philosophy of substance abuse, making GDC a good proxy for TAU in community programs.

Study therapists

Study therapists were 3 masters-level, licensed addiction counselors, and one doctoral-level psychologist. For GDC sessions, the psychologist did about half and one of the counselors did the remainder. S-BCT sessions were assigned to counselors on an alternating basis and as much as possible to avoid a patient’s S-BCT therapist being their GDC therapist. G-BCT sessions were mainly conducted by the same counselor – i.e. 85% of G-BCT sessions were run solely by this counselor, 7% were co-led with the first author for the first 9 group sessions, and 8% were led by another study therapist who filled in for vacation or sick days. Having a single ongoing GDC group and a single ongoing G-BCT group helped us meet the patient flow required for these rolling admission groups, and ensured continuity of group leadership. Having the therapist who provided the patient’s 12-step group counseling not be the same person who provided the patient’s couples counseling follows common practice. Therapists received a 1 day didactic training in S-BCT, G-BCT, and GDC. Therapists also received weekly supervision from the first author throughout the study.

Treatment fidelity

Sessions were audio-taped and rated to assess therapist adherence in delivering the manualized treatments. Items were rated on a 5-point scale (not at all to extensively). Adherence rating scales assessing fidelity to the GDC protocol were taken from Appendix C of the GDC manual (Daley et al., 2002). They consisted of 16 items evaluating the degree to which the therapist focused upon abstinence from substance use and the degree to which the therapist incorporated strategies for promoting abstinence which are consistent with a 12-step orientation. Adherence rating scales assessing fidelity to the S-BCT and G-BCT protocols included the same 10 items (O’Farrell, 2013) evaluating key aspects of the BCT protocol including review of promises, the trust discussion, and recovery contract as well as other activities scheduled to be completed per the BCT manual.

Sessions were randomly selected to be rated, and ratings were performed by 2 independent raters trained in delivering the treatments. Twenty each of the GDC, S-BCT and G-BCT sessions were rated, and 10 of each were independently coded by both raters. For sessions coded independently by both raters, level of agreement between the raters was moderately high, and the percentage of items that were rated within 1 point difference between raters were as follows: 90.6% GDC adherence, 83.3% S-BCT adherence, and 83.7% G-BCT adherence.

Adherence ratings were acceptable for each of the three session types. Mean ratings of adherence for each session type were as follows: GDC (M=4.68, SD = 0.49), S-BCT (M=4.42, SD=0.42), and G-BCT (M=4.50, SD=0.31), which are in the range between scale ratings of 4 = “considerably” and the maximum score of 5 = “extensively.” The S-BCT and G-BCT sessions did not differ significantly on mean adherence ratings, F(1,38) = 0.57, p = .46.

Analyses

Comparison of treatment conditions on substance-related and relationship satisfaction outcomes

For substance-related outcomes and relationship satisfaction outcomes, treatment condition marginal M and SE estimates were based upon results of generalized estimating equations (GEE). GEE has several advantages, including the ability to include cases with covariate-dependent missing data on the study outcome variables, the ability to manage variable non-independence, and the ability to directly include time as a predictor (Hall et al. 2001). Participants with at least one follow-up time period were included in these models. Following prior research (McCrady et al., 2009), an arcsine transformation was used to improve normality on PDA. Baseline days of non-study-based treatment was square root-transformed to improve normality. To separately test equivalency for patients versus partners on relationship outcomes, we computed different GEE models for patient- versus partner-reported DAS.

GEE analyses were conducted in steps. The first model included: a linear effect of time, the baseline outcome of interest, and baseline days of non-study-based treatment. We then compared relative model fits for dependent variable correlation matrices involving exchangeable, unstructured, and autoregressive structures and selected the model with the lowest Quasi Likelihood under Independence Model Criterion (QIC) value. Once the best-fitting correlational structure was determined, we added a quadratic effect of time to the model. If the addition of a quadratic effect of time was significant and improved the model fit according to the QIC, a quadratic effect of time was retained. Next, treatment condition was added followed by the treatment condition and time interaction. Finally, exploratory analyses were conducted by adding interactions between treatment condition and the baseline outcome of interest and between treatment condition and baseline days of non-study-based treatment. These interactions were nonsignificant in all analyses. Continuous predictor and outcome variables were z score-transformed to directly calculated effect size estimates and to reduce multicollinearity between main effects and interaction terms.

Equivalency tests

We used equivalency (non-inferiority) tests to investigate the hypotheses that S-BCT and G-BCT would demonstrate equivalency on study outcomes following treatment. S-BCT was coded as the reference group and G-BCT was coded as the comparison group. Following equations described in detail by Rogers, Howard, and Vessey (1993), we tested whether the mean for G-BCT was within 20% of the mean for S-BCT on a given study outcome during the follow-up period. This method involves computation of two one-sided z tests to determine whether the comparison group mean is significantly above the lower bound parameter and significantly below the upper bound parameter (i.e. +/− 10% of the M) of the reference group. To conclude statistical equivalency, both conditions must be satisfied - that is the comparison group mean must be found to be both significantly above the lower parameter (i.e., 10% below the M; lower bound critical z < −1.645) and significantly below the upper parameter (i.e., 10% above the M; upper bound critical z > 1.645) of the reference group. Following the recommendations by Rogers et al., we report the larger of the two p values for the upper and lower parameter z tests and conclude statistical equivalency only if the larger of the two p values is < .05.

Changes in outcomes from baseline

To examine improvement for patients in S-BCT and in G-BCT, we computed a series of paired sample t-tests to compare the baseline score from before treatment with each outcome score (post, 3-, 6-, 9-and 12-month follow-up) for each measure to determine whether patients in each condition showed significant improvement from baseline to that follow-up period. To control for type 1 error, we utilized a Bonferonni-corrected significance level of p < .01 for each comparison. Cohen’s (1988) d effect sizes were calculated for the paired sample comparisons, and we followed Cohen’s (1988) guidelines of d = .2 as small, d = .5 as medium, and d = .8 as large.

Power analysis

Assuming an equivalency parameter of +/−10% of the overall post- sample mean for PDA and α = .05, a statistical power analysis revealed that 94 participants per group would be required to achieve .80 power for determining statistical equivalency (Chow, Shao, & Wang, 2008).

Results

Patient Non-Study-Based Treatment

Treatment conditions did not differ on the amount of patient non-study-based treatment during the 60 days prior to joining the study, t(99) = 0.25, p = .80. Patients received an average of over 12 total days of detoxification, residential rehabilitation, and intensive outpatient treatment during the 60 days prior to the study (G-BCT M = 12.64, SD = 9.41; S-BCT M = 12.18, SD = 9.11).

Treatment Attendance and Satisfaction

Treatment conditions did not differ on number of study-based psychotherapy sessions that patients attended, t(99) = 0.13, p =.90. Patients in both conditions attended an average of 18 of 23 planned study therapy sessions: G-BCT M = 18.1, SD = 7.09; S-BCT M = 18.27, SD = 6.74. In addition, the conditions did not differ with regard to the number of patients completing at least 50% (i.e., 12 or more) of the prescribed study psychotherapy sessions, χ²(N = 101, 1) = 0.09, p = .76. In G-BCT, 80.0% completed at least 50% of the study sessions, and in S-BCT, 82.4% completed at least 50% of the study sessions. Finally, attendance at the S-BCT (M = 8.73, SD = 3.59) and at the G-BCT (M = 8.82, SD = 3.72) couple sessions did not differ significantly, t(99) = 0.13, p =.90.

According to the CSQ-8, patients and partners, on average, were very satisfied with their study-based treatment. The total CSQ-8 scores in G-BCT (patients: M = 28.55, SD = 3.87; partners: M = 27.05, SD = 4.64) and in S-BCT (patients: M = 28.53, SD = 3.81; partners: M = 28.63, SD = 4.01) did not differ significantly for either patients or partners (patients: t(90) = 0.03, p = .98; partners: t(89) = 1.75, p = .08). These CSQ-8 total scores were similar to large-scale samples of individuals seeking mental health care (Attkisson & Greenfield, 2004).

Primary Outcomes of Substance Use and Related Problems

Percent Days Abstinent

See Table 2 for descriptive statistics involving substance-related outcomes during the 12-month follow-up period. For the GEE models involving PDA, results supported a better fit for an autoregressive correlation model (QIC = 486.08) versus exchangeable (QIC = 487.30) or unstructured (QIC = 491.79). When a quadratic effect of time was added to the autoregressive model, there was a negligible improvement in the model fit (QIC = 485.86). However, the quadratic effect of time was nonsignificant (p =.061) and, therefore, not retained in subsequent models. The GEE analyses using an autoregressive structure and linear effect of time showed that while controlling for baseline PDA and baseline non-study-based treatment, neither the main effect of treatment condition nor the treatment by time interaction were significant (see Table 3). Despite the nonsignificant effect for treatment condition, the equivalency test failed to support the hypothesis that G-BCT was non-inferior to S-BCT on PDA, lower z = .097, p = .84. Taken together, these findings showed that although G-BCT and S-BCT were not significantly different, they were also not statistically equivalent on PDA.

Table 3.

Generalized Estimating Equation (GEE) Results for Outcomes During Follow-up

Percentage days abstinent (PDA) (N = 101)
Predictor	d	95% CI	χ2	p
Treatment condition	.010	−.372, .392	0.003	.959
Baseline PDA	.127	−.034, .288	2.382	.123
Baseline non-study SUD tx	.040	−.122, .201	0.231	.631
Time	−.018	−.099, .064	0.179	.673
Treatment x time	−.110	−.227, .007	3.382	.066

Inventory of Drug Use Consequences (InDUC) (N = 101)
Predictor	d	95% CI	χ2	p
Treatment condition	−.009	−.439, .260	0.252	.616
Baseline InDUC	.245	.078, .411	8.313	.004
Baseline non-study SUD tx	−.048	−.220, .124	0.294	.588
Time	−.081	−.149, −.013	5.421	.020
Treatment x time	.153	.054, .259	9.251	.002

Patient-reported Dyadic Adjustment Scale (DAS) (N = 98)
Predictor	d	95% CI	χ2	p
Treatment condition	.126	−.199, .452	0.578	.447
Baseline partner-reported DAS	.542	.405, .679	60.082	<.001
Baseline non-study SUD tx	.095	−.045, .236	1.767	.184
Time	.056	−.009, .121	2.896	.089
Treatment x time	−.150	−.244, −.057	9.874	.002

Partner-reported Dyadic Adjustment Scale (DAS) (N = 97)
Predictor	d	95% CI	χ2	p
Treatment condition	−.116	−.467, .235	0.419	.517
Baseline partner-reported DAS	.443	.307, .579	40.799	<.001
Baseline non-study SUD tx	.107	−.036, .250	2.140	.144
Linear time	−.246	−.493, .001	3.813	.051
Quadratic time	.042	.004, .080	4.686	.030
Treatment x linear time	−.125	−.236, −.013	4.806	.028

Note. Coding for treatment condition: 0 = group behavioral couples therapy, 1 = standard behavioral couples therapy. An autoregressive dependent variable correlation matrix was used for PDA and DAS, and an unstructured dependent variable correlation matrix was used for InDUC.

Substance-Related Problems

For the InDUC measure, an unstructured (QIC = 470.22) GEE correlational matrix fit better than an exchangeable (QIC = 472.54) or autoregressive (QIC = 471.13) matrix. The GEE analyses using an unstructured matrix did not show a significant main effect for treatment but did show a significant effect of time. The effect of time was qualified by a significant treatment by time interaction (see Table 3). Model-based GEE results probing the treatment by time interaction showed that the effect of treatment was non-significant at post (d = −.009, 95% CI = −.439, .260 p = .616), 3-month (d = .064, 95% CI = −.250, .378, p = .619) and 6-month (d = .217, 95% CI = −.091, .525, p = .167), but was significant at 9-month (d = .370, 95% CI = .038, .702, p = .029) and 12-month (d = .523, 95% CI = .142, .905, p = .007) follow-ups. These results were consistent with the observed means (see Table 2 and Figure 3) in showing that on the InDUC, there was an increasing advantage over the course of the follow-up period for S-BCT versus G-BCT, such that S-BCT had significantly fewer substance-related problems in the last 6 months of the 12-month follow-up period. Furthermore, S-BCT and G-BCT showed opposite trends over time, such that InDUC scores in S-BCT became less severe over the course of the 12-month follow-up period, while InDUC scores in G-BCT became more severe during the course of the 12-month follow-up period (see Table 2 and Figure 3). Turning to the equivalency test, findings did not support equivalency on the InDUC between S-BCT and G-BCT, upper z = −1.04, p = .85.

Figure 3.

Outcomes over time for S-BCT (standard behavioral couples therapy) and G-BCT (group behavioral couples therapy). Fig. 3a Percent Days Abstinent. Fig. 3b Negative Consequences (Inventory of Drug Use Consequences score). Fig. 3c DAS (Dyadic Adjustment Scale) patient report. Fig. 3d. DAS partner report.

Paired sample t-tests were computed to examine improvements in PDA and InDUC scores from baseline versus following treatment. As shown in Table 2, S-BCT and G-BCT both exhibited significant improvements in PDA and InDUC scores from baseline versus each time period following treatment. Both treatment conditions exhibited large Cohen’s d effect sizes when comparing baseline PDA and InDUC scores to time periods following treatment.

Secondary Outcome of Relationship Satisfaction

Patient Relationship Satisfaction

For the patient-reported DAS, an autoregressive GEE model (QIC = 341.76) produced better fit versus an exchangeable (QIC = 342.53) or unstructured (QIC = 365.49) model. The GEE analyses using an autoregressive correlation matrix did not produce a significant main effect for treatment condition, but there was a significant treatment condition by time interaction (see Table 3). Model-based GEE results probing the treatment by time interaction showed that the effect of treatment was non-significant at post (d = .126, 95% CI = −.199, .452, p = .447), 3-month (d = −.024, 95% CI = −.390, .261, p = .869) and 6-month (d = −.174, 95% CI = −.446, .098, p = .209), but was significant at 9-month (d = −.324, 95% CI = −.615, −.034, p = .028) and 12-month (d = −.475, 95% CI = −.809, −.140, p = .005) follow-ups. Furthermore, S-BCT and G-BCT showed opposite trends over time, such that patient-reported DAS scores in S-BCT increased over the course of the 12-month follow-up period, while DAS scores in G-BCT decreased during the course of the 12-month follow-up period (see Table 4 and Figure 3). Turning to the equivalency test, findings did not support equivalency on patient-reported DAS between S-BCT and G-BCT, lower z = 0.94, p = .83.

Table 4.

Relationship Outcomes Observed M (SD), Sample Size, and Effect Size Change from Baseline by Treatment Condition

	Patient-reported Dyadic Adjustment Scale (DAS)
Treatment	Baseline M (SD)	Post M (SD)	Post d	3-month M (SD)	3-month d	6-month M (SD)	6-month d	9-month M (SD)	9-month d	12-month M (SD)	12-month d
S-BCT	103.60 (18.14) n = 51	111.42a (17.52) n = 51	.50	113.04a (19.84) n = 48	.57	114.61a (20.55) n = 45	.57	114.37a,b (21.69) n = 46	.68	116.70a,b (20.26) n = 46	.82
G-BCT	97.81 (22.97) n = 50	111.18a (19.28) n = 47	.71	109.45a (21.69) n = 45	.55	105.25 (22.90) n = 44	.32	105.34 (24.32) n = 44	.28	104.24 (24.76) n = 43	.23

	Partner-reported Dyadic Adjustment Scale (DAS)
Treatment	Baseline M (SD)	Post M (SD)	Post d	3-month M (SD)	3-month d	6-month M (SD)	6-month d	9-month M (SD)	9-month d	12-month M (SD)	12-month d
S-BCT	90.12 (27.25) n = 51	110.15a (19.68) n = 49	.78	106.77a (24.53) n = 49	.66	108.83a,b (24.76) n = 47	.61	109.13a,b (24.52) n = 48	.68	112.79a,b (25.67) n = 46	.73
G-BCT	91.65 (23.06) n = 49	108.56a (16.37) n = 45	.87	101.98 (22.86) n = 43	.39	97.44 (28.95) n = 44	.20	97.41 (25.03) n = 44	.22	97.76 (28.74) n = 43	.20

Note. S-BCT = standard behavioral couples therapy. G-BCT = group behavioral couples therapy. Post = post-treatment.

^aThis score shows a significant (p < .01) improvement vs. baseline.

^bThis score shows a significantly more favorable outcome for S-BCT than G-BCT at this time period. Cohen’s d effect sizes guidelines are small = 0.2, medium = 0.5, large = 0.8

Paired sample t-tests showed significant, medium-to-large effect size differences on baseline patient DAS versus this measure at post-treatment. Throughout the follow-up period, S-BCT showed maintenance of significant medium-to-large effect size improvements on DAS scores versus baseline. In contrast, G-BCT showed a return to baseline on patient DAS scores, such that beginning at 6-month follow-up this measures was non-significant versus baseline.

Partner Relationship Satisfaction

For the partner-reported DAS, an autoregressive GEE model (QIC = 398.36) produced better fit versus an unstructured (QIC = 451.71) or exchangeable (QIC = 401.03) model. The addition of a quadratic effect of time to the autoregressive model improved the model fit (QIC = 386.69), and the quadratic effect of time was significant (p = .03). Therefore, a linear plus quadratic effect of time and autoregressive correlation structure were used in subsequent GEE models that included the effect of treatment condition. GEE analyses did not produce a significant main effect for treatment condition, but there was a significant treatment condition by linear time interaction (see Table 3). Model-based GEE results probing the treatment by linear time interaction showed that the effect of treatment was non-significant at post (d = −.116, 95% CI = −.467, .235, p = .517) and 3-month (d = −.241, 95% CI = −.536, .055, p = .11), but was significant at 6-month (d = −.365, 95% CI = −.641, −.090, p = .009), 9-month (d = −.490, 95% CI = −.790, −.190, p = .001), and 12-month (d = −.615, 95% CI = −.973, −.256, p = .001) follow-ups. Furthermore, S-BCT and G-BCT showed differing trends over time. Both treatment conditions showed an initial decline in partner-reported DAS from post to 3-month follow-up. However, in S-BCT partner-reported DAS then rebounded and increased from 3-month to 12-month follow-up, while in G-BCT partner-reported DAS scores further declined from 3-month to 6-month follow-up and then failed to rebound (see Table 4 and Figure 3). Turning to the equivalency test, findings did not support equivalency on partner-reported DAS between S-BCT and G-BCT, upper z = −0.16, p = .56.

Paired sample t-tests showed significant, medium-to-large effect size differences on baseline partner DAS versus this measure at post-treatment. Throughout the follow-up period, S-BCT showed maintenance of significant medium-to-large effect size improvements on DAS scores versus baseline. In contrast, G-BCT showed a return to baseline on partner DAS scores, such that beginning at 3-month follow-up this measure was non-significant versus baseline.

Discussion

This randomized clinical trial compared G-BCT plus IBT with S-BCT plus IBT over a 1-year follow-up period. We tested the prediction that substance and relationship outcomes for G-BCT plus IBT would be equivalent (i.e., non-inferior) to outcomes for S-BCT plus IBT. However, results found no support for this prediction in the analyses testing statistical equivalency. Further, substance and relationship out-comes were significantly worse for G-BCT than S-BCT in the last 6–9 months of the 12-month follow-up period, because G-BCT deteriorated and S-BCT maintained gains during follow-up. ²

On primary outcomes of substance-related problems and days abstinent, contrary to predictions, G-BCT and S-BCT did not have statistically equivalent outcomes. When improvements from baseline on substance-related problems and PDA were considered, both G-BCT and S-BCT were significantly improved at post and all follow-ups, with the extent of improvement reflected by a large effect size. Despite this similar extent of improvement, trends over time on substance outcomes after treatment diverged for G-BCT and S-BCT. This was most pronounced for substance-related problems. Problems increased over time for G-BCT and decreased for S-BCT, such that G-BCT had significantly greater substance problems in the last 6 months of the 12-month follow-up period. For PDA there was a similar pattern of results. Days abstinent decreased over time in the year after treatment for G-BCT and increased or maintained for S-BCT. However, for PDA this result only approached significance.

On secondary outcomes of relationship satisfaction for the patient and partner, contrary to predictions, G-BCT and S-BCT did not have statistically equivalent outcomes. Instead, worse outcomes over time occurred for G-BCT. Relationship satisfaction improved significantly from before to immediately post-treatment in both conditions. For S-BCT, these post-treatment outcomes were maintained or enhanced over time, remaining significantly improved through 12-month follow-up. However, for G-BCT, relationship outcomes deteriorated after treatment such that G-BCT was no longer significantly improved after the immediately post-treatment period for partners or after 3-month follow-up for patients. Further, when treatment conditions were compared, G-BCT had significantly lower relationship satisfaction than S-BCT in the last 6 months (for patients) to 9 months (for partners) of the 12-month follow-up period.

Why were results for the newer rolling admission group format for BCT not equivalent to standard conjoint BCT? In fact why did G-BCT have significantly worse outcomes than S-BCT? We can consider possible explanations. First, possibly S-BCT couples got more intensive and individualized training in BCT-related behaviors designed to support abstinence and improve communication. They may have learned these skills better and used them more consistently during treatment and also after treatment ended. This may have led to more durable maintenance of behavior change during the follow-up period. This suggests that future studies could examine whether the extent of couples’ use of BCT-targeted behaviors during treatment and in the year after treatment mediates the superior outcomes observed for S-BCT than G-BCT. Second, S-BCT may have made it easier for therapists to provide effective crisis intervention for drinking and relationship crises because they did not have to attend to other couples as they did in G-BCT. This may have increased a couple’s self-efficacy about their ability to manage their problems, and this may have led them to get “back on track” before drinking or relationship problems worsened substantially during the follow-up period. Third, in S-BCT couples may have shared more of their problems and received a more in-depth response. In S-BCT, the therapist had more time to devote to one rather than multiple couples, and the couple may have been more comfortable revealing themselves, especially sensitive issues, to the therapist alone as compared with multiple other couples. Fourth, this was the first study of a rolling admission BCT group. Despite that both G-BCT and S-BCT had similarly high ratings of therapist adherence and participant satisfaction, therapists had more experience delivering S-BCT and this may have played a role in the worse G-BCT outcomes. Finally, is it possible that the rolling admission approach reduced the efficacy of G-BCT? S-BCT had 4 modules in the following sequence: (1) introductory sessions to negotiate a Recovery Contract to reinforce sobriety, (2) positive feelings and activities, (3) basic communication, and (4) advanced communication and conflict resolution skills. G-BCT couples had the same introductory sessions and the same modules 2–4 but not necessarily in the same sequence. For example with rolling admissions, depending on when they joined the group, a couple might start G-BCT with advanced communication before basic communication or positive activities. Although we had been concerned that this “out of sequence” material might be problematic, our impressions were that this did not present difficulties after all. The fact that both G-BCT and S-BCT had the same introductory sessions and had similar ongoing session structure (e.g., review Recovery Contract and sobriety in past week, report on homework, followed by new material in second half of session), seemed to have overcome the relatively minor difference in sequence of content experienced by some G-BCT couples. Future studies of mechanism of action and necessary and active ingredients of BCT could help explain why initial G-BCT gains eroded over time.

Implications of the present study should be considered. First, clinical outcomes favor S-BCT over G-BCT. If results of the present study showing worse outcomes for G-BCT than S-BCT are replicated in future research, it would suggest that AUD treatment programs should consider using S-BCT rather than G-BCT. BCT may be an area of AUD treatment for which the dominant rolling admission group counseling approach does not produce optimal outcomes, and a more individualized one couple at a time method is preferable. Second, logistical considerations also may favor S-BCT over G-BCT, and when considered along with better outcomes, add impetus for increased utilization of S-BCT rather than G-BCT. Logistical challenges of a G-BCT rolling admission couples group include the constant need to replenish new couples as other couples graduate. This is especially a concern for smaller clinics that do not have enough patients with partners to maintain an ongoing couples group. Finally, economic considerations may favor S-BCT over the long run. Although S-BCT may be more expensive than G-BCT in the short run due to costs associated with therapist time, in the long run, there may be greater cost savings for S-BCT due to its more durable effects with regard to substance use and relationship outcomes. If future economic studies showed that S-BCT was more cost effective over the long run, this could influence policy makers’ decisions about how to allocate scarce resources despite the greater upfront cost of providing treatment in this format.³

This study had multiple strengths. These included the high treatment fidelity ratings achieved by the study therapists, the high session attendance rates and therapy satisfaction ratings in both S-BCT and G-BCT suggesting they were equally credible and satisfying treatments, use of accepted outcome measures and collateral reports to reduce under-reporting, use of an intent-to-treat sample and accepted statistical analyses, low degree of missing data, and low attrition rates. Further, as would be done in many SUD treatment programs, S-BCT and G-BCT were delivered in combination with a 12-step oriented group counseling intervention that has been shown to be efficacious in its own right (Crits-Christoph et al., 1999).

Limitations of the study should also be noted. First, the present study did not include a comparison group in which patients received IBT only without any BCT couple sessions. Most prior studies have compared S-BCT to IBT and found better outcomes for S-BCT (Meis et al., 2013; Powers et al., 2008). The lack of an IBT-only comparison group in the present study precluded testing the efficacy of G-BCT versus IBT and replicating S-BCT outcomes.

Second, the potential role of regression to the mean in the current results should be considered, given the use of the most extreme baseline report of substance related outcomes along with the absence of a control group. Regression to the mean could account in part for the robust improvements on substance related outcomes observed in both S-BCT and G-BCT.

Third, research assistants who collected study outcome data were not blind to treatment conditions of study participants.

Fourth, patients who received G-BCT sessions received a greater dose of therapy than those who received S-BCT sessions (i.e., 2 60-min and 9 90-min sessions of G-BCT versus 11 60-min sessions of S-BCT). This differing session length for S-BCT and G-BCT was chosen in the study design to reflect the way each intervention would likely be delivered in routine practice settings. Notably, this difference might seem to give the G-BCT group an advantage (e.g., more therapist contact). However, G-BCT results did not reflect this advantage. Still the possibility that differential session length played a role in the present study findings cannot be ruled out.

Fifth, the fact that a single provider delivered most of the G-BCT sessions raises the concern that characteristics of this one therapist might have caused the poorer BCT outcomes rather than the different characteristics of the G-BCT vs. S-BCT formats. However, adherence, satisfaction and attendance scores were all high and did not differ between G-BCT and S-BCT. These scores suggest that the poorer results found for G-BCT were not the result of an under-performing therapist.

Sixth, the study was somewhat underpowered to be able to test for statistical equivalency between S-BCT and G-BCT. However, the fact that S-BCT was significantly superior to G-BCT during the follow-up period suggests that a larger sample would be unlikely to produce equivalency findings between S-BCT and G-BCT.

Seventh, the representativeness of the current sample should be considered. The present study may have had a fairly select sample due to high study refusal rates and scheduling constraints. Sixty-five percent of potentially eligible patients refused the study with the most frequent reason for refusal being lack of time on the part of the patient or the partner. In terms of scheduling, GDC for patients was offered on one evening, and G-BCT for couples was offered on a second evening, and both patient and partner had to be available that evening in case they were randomized to G-BCT. Unfortunately, we do not have information on characteristics of refusing patients so we cannot determine if they differed from those in the study sample and the nature of any selection bias that may have occurred. Therefore the generalizability of the present results may be somewhat limited, and high refusal rates have been reported in other studies. As noted by McCrady and colleagues (McCrady, Epstein, Cook, Jensen, & Ladd, 2011) in their study of patient-focused barriers to BCT, most BCT studies have a substantial refusal rate because of partner reluctance, patients not wanting their partner involved, scheduling challenges, and other concerns. Thus, it is appropriate to qualify reports of positive outcomes of BCT, both in the present study and in the broader literature, by noting that the findings apply to couples with a willing partner that the patient would like to have involved in the treatment. Given high refusal rates in BCT studies, future work should examine further the nature of and remedies for patient-focused barriers to BCT (McCrady et al., 2011).

Finally, the present results may not generalize to more ethnically diverse samples, same sex couples, couples in which both partners have a substance use disorder, and primary drug abuse patients – all of which were not included in the present study sample.

In conclusion, this was the first study to test a new rolling admission couples group format for BCT. Results showed that both S-BCT and G-BCT were associated with statistically significant and large improvements in substance-related outcomes, but S-BCT had significantly better maintenance of treatment gains following treatment. Further, only those who received S-BCT exhibited lasting improvements in relationship satisfaction following treatment. Results from this study are consistent with prior studies which show that S-BCT is more efficacious than other credible psychotherapy comparators. These findings suggest that in order to achieve maximum durable clinical benefits, clinicians and program administrators may want to consider the standard, one couple at a time, format in delivering BCT.

Public Health Significance.

This study showed that when compared to a new rolling admission group format, a standard one couple at a time format for delivering behavioral couples therapy produced more lasting benefits for reducing substance use problems and improving relationships.