Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Feb 23.
Published in final edited form as: J Consult Clin Psychol. 2020 Dec;88(12):1119–1132. doi: 10.1037/ccp0000615

Patterns of Transitions between Relapse to and Remission from Heavy Drinking over the First Year after Outpatient Alcohol Treatment and Their Relation to Long-Term Outcomes

Stephen A Maisto 1, Kevin A Hallgren 2, Corey Roos 3, Julia Swan 4, Katie Witkiewitz 4
PMCID: PMC7900838  NIHMSID: NIHMS1664580  PMID: 33370135

Abstract

Objectives:

Studying clinical course after alcohol use disorder (AUD) treatment is central to understanding longer-term recovery. This study’s two main objectives were to (1) replicate a recent study that identified heterogeneity in patterns of remission from/relapse to heavy drinking during the first year after outpatient treatment in an independent data set, and (2) extend these recent findings by testing associations between patterns of remission/relapse and long-term alcohol-related and functioning outcomes.

Method:

Latent profile analyses were conducted using data from Project MATCH (N=952; mean age=38.9; 72.3% female) and COMBINE (N=1,383; mean age=44.4; 69.1% male). Transitions between heavy and non-heavy drinking within consecutive, 2-week periods over a one-year post-treatment period were characterized for each participant. From this, latent profiles were identified based on participants’ initial 2-week heavy drinking status, the number of observed transitions between 2-week periods of relapse and remission, and the average duration of observed remission/relapse episodes.

Results:

In both MATCH and COMBINE we identified 6 profiles: 1) “continuous remission,” 25.3% of COMBINE sample/ 25.3% of MATCH sample; 2) “transition to remission”, 19.6%/9.6%; 3) “few long transitions,” 15.9%/33.7%; 4) “many short transitions,”13.2%/13.6%; 5) “transition to relapse,” 7.2%/7.1%; and 6) “continuous relapse,” 18.8%/10.5%. Profiles 1 and 2 had the best long-term outcomes, profiles 5 and 6 had the worst, and profiles 3 and 4 fell between these groups.

Conclusions:

That many individuals can remit from heavy drinking following one or more relapses to heavy drinking may be of direct interest to individuals in recovery from AUD.

Keywords: alcohol use disorder, recovery, relapse, remission, treatment

Clinical course of alcohol use disorder (AUD) is the progression of change in AUD “symptoms” following the initiation of formal treatment or self-initiated behavior change (Maisto et al., 2014, 2018). Because it is central to understanding maintenance of change in AUD, AUD clinical course is clinically significant and has been studied for decades (Brownell et al., 1986; Kelly et al., 2019; Lowman et al., 1996; MacKillop, 2020; Moos & Moos, 2005, 2006, 2007). This and related research have shown across a range of samples of individuals presenting for AUD treatment that AUD clinical course is heterogeneous across individuals, which has led to current conceptualizations that AUD clinical course has major markers (response, remission, relapse, recovery, and recurrence; after Frank et al.’s, 1991, identification of “change points” in the clinical course of depression) that are typical points in the change process. Nevertheless, Maisto et al.’s (2016) systematic review of studies published 2010–2015 showed that remission and relapse were overwhelmingly viewed as static outcomes rather than as part of a change process, a finding that has considerable implications for both clinical practice and for advancing knowledge about AUD clinical course.

Accordingly, Maisto et al. (2018) reported one of the few studies of AUD clinical course in which indicators of relapse and remission were viewed as markers of a change process, rather than as static outcomes, in alignment with theoretical models of and data on AUD clinical course as cited earlier. Maisto et al. conducted secondary data analyses of Project Matching Alcoholism Treatment to Client Heterogeneity (MATCH; Project MATCH Research Group, 1997) outpatient subsample data to investigate AUD remission and relapse as parts of a process of behavior change. Two-week periods of relapse (i.e., any observed heavy drinking days) and remission (i.e., no observed heavy drinking days) were identified for each participant and transitions between these states were coded to investigate the relation between shorter-term remission and relapse events to longer-term course at one- and three-year post-treatment conclusion.

Maisto et al. (2018) identified six latent profiles based on the frequencies of transitions between two-week relapse and remission periods the durations of relapse and remission episodes over the first year following outpatient AUD treatment. Profiles identified in this study included: (1) continuous remission from heavy drinking, (2) approximately one transition to remission from heavy drinking and staying in remission following this transition, (3) a few, longer transitions between heavy drinking and no heavy drinking, with each transition last an average of about 2 months, (4) numerous shorter transitions, with each episode lasting an average of about 1 month, (5) about one transition to relapse to heavy drinking and staying in relapse after this transition, and (6) continuous relapse. Profile membership was differentially predictive of alcohol use (percentage days drinking (PDD), percentage of days heavy drinking (PHDD), drinks per drinking day (DDD)), drinking consequences, all in the last 90 days, and non-alcohol (depression in the last 2 weeks, and psychosocial functioning since the last interview) outcomes at both one and three years following treatment. One- and three-year outcomes had gradated associations with the six profiles: at the one-year follow-up, profiles 1 and 2 had the best outcomes and did not differ from each other, profiles 3 and 4 generally had better outcomes than profiles 5 and 6 and worse outcomes than profiles 1 and 2 on most outcomes. The three-year pattern of results was highly similar to that found for year one. Maisto et al. interpreted these findings as reaffirming the heterogeneity of AUD clinical course and the importance of viewing remission from and relapse to heavy drinking as dynamic aspects of the AUD clinical course, as opposed to static outcomes. Consistent with other recent work (Wilson et al., 2016), these data suggest that clinicians and researchers be advised not to refer to “any heavy drinking” following treatment as treatment failure. Rather, the focus would seem to be better placed on whether individuals can transition out of relapse to heavy drinking and sustain remission, which is associated with better recovery of psychosocial functioning. On the other hand, for individuals who do relapse, the focus should be on reducing the duration and/or frequency of relapse episodes.

A major limitation of this prior work is that the profiles derived were based on a single sample, and although it was an unusually large one that represented multiple treatment sites, the generalizability and replicability of the pattern of results is an empirical question. In addition, it would be of substantial interest to determine if the predictive power of likely profile membership extended beyond the three-year post-treatment mark.

The purpose of the present study was to replicate the findings of Maisto et al. (2018) using data from a new sample – the Combined Pharmacotherapies and Behavioral Interventions for Alcohol Dependence (COMBINE) study – in order to test whether the latent profiles derived in previous work would replicate in an entirely new sample of AUD treatment outpatients. Our study also aimed to extend the previous findings by modeling associations of latent profiles with longer-term outcomes beyond three years post-treatment in both the Project MATCH (10 years) and COMBINE study (7–9 years) samples. We also modeled their associations with additional non-alcohol use outcomes not included in Maisto et al., including AUD and mental health service utilization, recovery efforts, physical health, and subsequent quit attempts, in alignment with a recent definition of recovery as “…a dynamic process of change characterized by improvements in health and social functioning, as well as increases in well-being and purpose in life” (Witkiewitz et al., in press). If previous findings are replicated in the new sample (e.g., similar number and patterns of latent profiles; similar associations with longer-term drinking and non-drinking outcomes), it would support the validity of our conclusions regarding transitions between relapse and remission as dynamic processes within the AUD clinical course and would support the generalizability and robustness of previous findings. Likewise, if the associations between latent profiles and other long-term outcomes extend to 7- to 9-year or 10-year post-treatment outcomes, or to additional non-drinking outcomes, it would further affirm the importance of relapse and remission patterns in overall long-term health.

Method

Participants and procedures

We used data from the COMBINE Study (Anton et al., 2006) and the outpatient sample of Project MATCH (Project MATCH Research Group, 1997), both multi-site randomized clinical trials, in the current analyses. Participants in both studies met criteria for alcohol abuse or dependence (COMBINE required alcohol dependence; MATCH included individuals who met criteria for alcohol abuse or dependence), received outpatient treatment, and were followed for up to three years post-treatment (COMBINE: n=694; MATCH: n=806), 7–9 years post-treatment (COMBINE: n=133), and 10 years post-treatment (MATCH: n=150). Exclusion criteria for both studies included a current drug use disorder (other than nicotine or cannabis) or a psychiatric disorder requiring medication. COMBINE had additional exclusion criteria based on medication contraindications, and both studies excluded individuals who were not medically stable.

COMBINE.

Participants recruited for the COMBINE Study (n=1383) met criteria for alcohol dependence based on the DSM (fourth edition)(American Psychiatric Association, 1994). Demographics of the sample were: 69.1% male, 76.8% non‐Hispanic White, 46.3% married, and mean age 44.4 (SD = 10.2). Based on the Form 90 data at baseline, participants had an average of 12.96 drinks per drinking day (SD = 8.07), 78.59% drinking days (SD = 22.50), and 70.52% heavy drinking days (SD = 26.57). Participants were randomized to receive 16 weeks of treatment in one of nine treatment conditions based on a 2 × 2 × 2 + 1 design. The 2 × 2 × 2 conditions received (1) active naltrexone (100 mg/day) or placebo naltrexone, (2) active acamprosate (3000 mg/day) or placebo acamprosate, and (3) medication management (MM) or combined behavioral intervention (CBI) with MM. One additional condition (+1) received CBI only and no pills.

Of the 1383 patients recruited for the COMBINE trial, 874 patients from nine sites (n = 1144; 76.4%) consented to the three-year long-term follow-up (Zarkin et al., 2008, 2010) and 694 (79.4%) provided data at the three-year assessment. An additional 198 individuals from four of the eleven sites agreed to enroll in a long-term follow-up to occur at 7 to 9 years following treatment, and 133 (67%) completed that long-term follow-up. Those who completed the three-year follow-up had significantly higher baseline alcohol dependence severity, as measured by the Alcohol Dependence Scale (Skinner & Horn, 1984) (t (1373)=2.35, p=0.02), and those who completed the 7- to 9-year follow-up were significantly older (t (1381)=−2.34, p=0.02), had significantly fewer drinks per drinking day (t (1381)=2.61, p=0.01), and had fewer alcohol-related consequences at baseline (t (1378)=2.49, p=0.01).

Project MATCH.

Participants recruited for the outpatient sample of Project MATCH (n=952) met the criteria for alcohol abuse (4.6%) or alcohol dependence (95.4%) based on the DSM (third edition-revised)(American Psychiatric Association, 1987). Demographics of the outpatient sample were: 72.3% male, 80% non‐Hispanic White, 35.5% married, and mean age 38.9 (SD = 10.7). Based on the Form 90 data at baseline, participants had an average of 13.62 drinks per drinking day (SD = 8.76), 65.77% drinking days (SD = 29.83), and 58.33% heavy drinking days (SD = 30.78). Participants were randomized to receive 12 weeks of treatment with Cognitive Behavioral Therapy (Kadden et al., 1995), Motivational Enhancement Therapy (Miller et al., 1994), or Twelve-Step Facilitation (Nowinski et al., 1995).

Of the 952 patients recruited for MATCH, 806 patients (85%) completed the three-year follow-up assessment. A portion of the Albuquerque outpatient sample who were reconsented at 3 years (n=193) and still alive at 10 years following treatment provided follow-up data at the 10-year follow-up (n =150; 78% retention). There were no significant differences between study participants with and without 10-year data (Pagano et al., 2013).

Measures

Alcohol consumption.

Alcohol consumption was assessed using the Form 90 (Miller, 1996), a calendar-based, self-report method to determine daily alcohol use. The Form 90 was administered at each post-treatment follow-up visit to quantify times of relapse and remission, as further illustrated in the Statistical Analyses section below. For COMBINE, this would include 3- and 7- to 9-year follow-ups post-treatment, and for Project MATCH, this would include 3- and 10-year follow-ups post-treatment. For the current analyses, we utilized the summary drinking measures at follow-ups including drinks per drinking day (DDD; i.e. the average number of drinks per drinking day), percent drinking days (PDD; i.e. the percentage of days in which alcohol was consumed), and percent heavy drinking days (PHDD; i.e. percent of days with at least 4/5 drinks for women/men). Previous studies have demonstrated the reliability and accuracy of the Form 90 (e.g. Witkiewitz, Finney, Harris, Kivlahan, & Kranzler, 2015a, 2015b).

Alcohol-related consequences.

The Drinker Inventory of Consequences (DrInC; Miller, Tonigan, & Longabaugh, 1995) was used during the 10-year assessment in Project MATCH to evaluate self-reported alcohol-related negative consequences. Participants determined the frequency of 45 alcohol-related consequences in the past year on a 4-point scale (1 = never, 4 = daily or almost daily). Internal consistency reliability in Project MATCH was α=0.98.

Mental health.

The 12-item short-form health survey (SF-12; Ware, Kosinski, & Keller, 1996) was used to assess mental health presently and in the past four weeks in the COMBINE study at the 3-year follow-up. The mental health composite score asks how affective symptoms, such as feeling depressed or anxious, have affected functioning in daily life. The SF-12 has previously been shown to have good internal consistency (e.g. Huo, Guo, Shenkman, & Muller, 2018), and in COMBINE the SF-12 had an internal consistency reliability of α=0.91.

The Beck Depression Inventory (BDI, Beck, Steer, & Carbin, 1988) is a self-report scale that measures depressive symptoms that was utilized at the 10-year follow-up in Project MATCH. Participants respond to 21 items on a scale from 0 to 3 (e.g. 0 = “I don’t feel particularly guilty”, 3 = “I feel guilty all the time”) corresponding to the degree to which each statement matches how they have been feeling during the past two weeks. Participants are classified based on total scores with 14 or greater defined as “mild”, 20 or greater as “moderate”, and 29 or greater as “severe”. In Project MATCH, internal consistency reliability was α=0.91.

The State-Trait Anger Expression Inventory (STAXI; Spielberger, 1988) is a 44-item, self-reported scale measuring state anger, trait anger, and anger expression that was used in Project MATCH at the 10-year follow-up. Participants respond on a four-point frequency scale (1 = almost never, 4 = almost always), and internal consistency reliability in Project MATCH at the 10-year assessment was α=0.87. A single question asking about seeking treatment for mental or substance-related problems in the past 12 months was used as an indicator of mental health service utilization during the COMBINE 7- to 9-year follow-up.

Quality of life.

The World Health Organization Quality of Life Scale-BREF (WHOQOL-BREF; WHOQOL Group, 1998) was used in COMBINE at the 3-year assessment to determine participants’ perceived quality of life and health in the past four weeks. This questionnaire consists of 26 items with varying 5-point scales (1 = Very poor; very dissatisfied; not at all to 5 = Very good; very satisfied; an extreme amount) depending on the question. In COMBINE, the WHOQOL-BREF had an internal consistency reliability of α=0.91.

The Purpose in Life questionnaire (PIL; Crumbaugh & Maholick, 1964) was utilized in the 10-year assessment for Project MATCH. This 20-item self-report questionnaire with 7-point unique Likert scales for each item (e.g. 1 = “often wonder why I exist”, 4 = neutral, 7 = “always see a reason for my being here”). Participants are instructed to choose the response that is most true for them and to avoid using the neutral rating. The internal consistency reliability was α=0.92 in Project MATCH.

Physical health.

The SF-12 (Ware et al., 1996) was also used to assess physical health in the COMBINE study at the 3-year follow-up. The physical health composite score assesses how physical health, including pain, impacts daily functioning. Internal consistency reliability was α=0.91. In the COMBINE study during the 7- to 9-year assessment, a single question of general health (“In general, how you would rate your current health?”) was utilized in analyses. Response options for this question were on a 4-point scale (1 = excellent, 2 = good, 3 = fair, and 4 = poor).

Use of 12-step programs.

At the 10-year assessment for Project MATCH, AA-specific recovery efforts were measured using The General Alcoholics Anonymous Tools of Recovery (GAATOR 2.1; Montgomery, Miller, & Scott Tonigan, 1995) is a 25-item self-report questionnaire assessing how effectively participants have utilized the 12 steps of Alcoholics Anonymous and other 12-step programs (e.g. Narcotics Anonymous) in the past 90 days. Four-point responses for each item range from “Definitely False” to “Definitely True” with items such as “I have believed that my recovery could only come from a power greater than myself” and “I have made direct amends to those whom I had harmed.” The internal consistency reliability was α=0.96.

To measure efforts to modify drinking, in the COMBINE study during the 7- to 9-year follow-up, a single question of whether the participant attempted to quit or reduce drinking in the past 12 months was used in analyses. Unfortunately, the COMBINE and MATCH datasets did not include additional measurement of other alcohol interventions (e.g., types, durations, etc.) that were received during the follow-up periods. However, both the GAATOR and attempts to quit or reduce drinking are presumed to reflect different aspects of efforts to recover.

Statistical Analyses

COMBINE Latent Profile Analyses.

Consistent with prior analyses of the MATCH data (Maisto et al., 2018), the daily drinking data from the yearlong post-treatment period of COMBINE (starting 112 days/16 weeks and ending 476 days/68 weeks after the start of treatment) were classified as either heavy (≥ 4/5 drinks for women/men) or non-heavy drinking days (no drinking or drinking <4/<5 drinks for women/men) and were then combined into 26 sequential 14-day periods with binary indicators of any heavy drinking or no heavy drinking during each period1. We started with the first 14-day period following treatment to define initial post-treatment status as heavy drinking or no heavy drinking. Following this initial period, the observed transitions to remission from heavy drinking were defined by having a 14-day period with any heavy drinking that was followed by a 14-day period with no heavy drinking. Observed transitions to relapse to heavy drinking were defined by 14-day period with no heavy drinking that was followed by a 14-day period with any heavy drinking. Durations of episodes of remission from heavy drinking were calculated based on the duration of consecutive 14-day periods in remission with no heavy drinking. Durations of episodes of relapse to heavy drinking were calculated based on the duration of consecutive 14-day periods with at least one heavy drinking day.

We then used latent profile analyses with a robust maximum likelihood estimator in Mplus version 8.4 (L. K. Muthén & Muthén, 2019), adjusting the standard errors for clustering within site, to examine AUD clinical course based on four indicators: (1) heavy drinking status in the first two weeks following treatment, (2) number of observed transitions between remission from and relapse to heavy drinking, (3) duration of episodes of remission from heavy drinking, and (4) duration of episodes of relapse to heavy drinking. Individuals with no heavy drinking during the first two weeks following treatment who never relapsed to heavy drinking during the 1-year follow-up were considered in remission. Individuals with heavy drinking during the first two weeks following treatment who never transitioned to remission were considered relapsed. These remission and relapse profiles were known (individuals who were in remission or relapsed), and constraints were placed on the data to ensure these two profiles were part of the final latent profile solution. Because we were interested in studying the replication of prior work (Maisto et al., 2018), we started with a 6-profile latent profile solution using confirmatory latent profile analysis (Finch & Bronk, 2011), where mean differences in outcomes by profiles were constrained to reflect the patterns of means observed in Maisto et al (2018). We also examined the Bayesian Information Criterion (BIC) and sample-size-adjusted BIC (aBIC) to assess whether the 6-profile model provided a reasonable fit to the data, as compared to 5- and 7-profile models. Likelihood ratio tests (including the Lo Mendell Rubin (LMR) and Bootstrapped Likelihood Ratio test (BLRT)) are commonly employed to determine the number of classes (Nylund et al., 2007), but the BLRT is not available for complex sampling designs with clustering by site. A non-significant LMR indicates that adding an additional profile (e.g., 7 profiles) does not significantly improve model fit. Lower BIC and aBIC indicates a better fitting model. We also examined classification precision (defined by relative entropy, range from 0 to 1, with values near 1 indicating greater classification precision) to determine how well the final latent profile solution classified individuals into profiles (Nylund et al., 2007). Consistent with Maisto et al (2018), we did not include any covariates (including treatment group membership) as predictors of latent profiles. Thus, profiles were not influenced by any pre-treatment characteristics or treatment assignment, and were estimated to only reflect post-treatment patterns of transitioining between remission from and relapse to heavy drinking.

COMBINE and MATCH Distal Outcome Analyses.

Once the final latent profile solution was selected for COMBINE, we used distal outcome analysis via the manual Bolck, Croon, and Hagenaars (BCH) approach (Bolck et al., 2004; Nylund-Gibson et al., 2019) to examine profile differences in 1-, 3-, and 7- to 9-year post-treatment outcomes in COMBINE. The BCH approach accommodates uncertainty in profile assignment in estimating the differences in outcomes across latent profiles and has been shown to produce less biased estimates than alternative approaches (Asparouhov & Muthén, 2014; Bray et al., 2015). Next, we also used the manual BCH approach to test the profile differences in 10-year post-treatment outcomes for the 6-profile solution identified in prior analyses of Project MATCH data (Maisto et al., 2018). For these models, we used the final latent profile model from Maisto et al (2018) estimated in the Project MATCH data, and then added distal outcomes at the 10-year follow-up of Project MATCH. Distal outcome analysis provides an estimate of the mean differences in outcomes across profiles using a Wald χ2 test. We were particularly interested in effect sizes for differences in distal outcomes across profiles, thus we report Cohen’s d effect size values for the standardized differences between latent profiles.

Results

COMBINE Latent Profile Analyses

Results of the confirmatory latent profile analyses indicated that a 6-profile model, with specification based on the 6-profile model previously identified in the Project MATCH data (reported in top of Table 1; Maisto et al., 2018), provided a good balance of parsimony and model fit, with lower BIC/aBIC than the 5-profile model (see Supplementary Table 1), and the 7-profile model did not fit significantly better than the 6-profile model based on the LMR (p=0.24).2 The 6-profile model also provided excellent entropy (0.974) and yielded interpretable latent profiles that could be interpreted similarly to prior analyses and that mostly had similar proportions of participants in each profile (see percentages below, MATCH values reflect the proportion of participants assigned to each profile from Maisto et al (2018) for comparison, COMBINE values reflect new results). As shown in Table 1, we identified the following profiles of relapse and remission over the 1-year period following treatment (prevalence rates are based on estimated posterior probabilities of profile membership): (1) continuous remission from heavy drinking (MATCH: 25.5%/COMBINE: 25.3%); (2) approximately one observed transition to remission from heavy drinking and mostly staying in remission after this transition (MATCH: 9.6%/COMBINE: 19.6%); (3) a few long transitions with a mix of remission from and relapse to heavy drinking (MATCH: 33.7%/COMBINE: 15.9%); (4) numerous transitions with each heavy drinking remission or relapse episode lasting approximately 1 month on average (MATCH: 13.6%/COMBINE: 13.2%); (5) approximately one observed transition to relapse heavy drinking and staying in relapse after this transition (MATCH: 7.1%/COMBINE: 7.2%); and (6) continuous relapse (MATCH: 10.5%/COMBINE: 18.8%). As shown in Table 1, the two studies differed with respect to the duration of heavy drinking remission and relapse episodes in profiles 2 through 5, where episodes of remission from heavy drinking were generally shorter in COMBINE and episodes of relapse to heavy drinking were generally longer. In COMBINE profile 3 (few long transitions), the duration of episodes of relapse to heavy drinking was nearly twice as long as the duration of episodes of remission from heavy drinking, and more participants were classified in the continuous relapse profile.

Table 1.

Latent Profile Prevalence and Indicator Proportions and Means (Standard Errors) by Relapse Course Latent Profiles

Project MATCH (n=877)
Indicators Profile 1: Continuous remission (25.5%) Profile 2: Transition to remission (9.6%) Profile 3: Few long transitions (33.7%) Profile 4: Numerous short transitions (13.6%) Profile 5: Transition to relapse (7.1%) Profile 6: Continuous relapse (10.5%)
Initial heavy drinking Proportion 0.00 0.42 0.46 0.39 0.49 1.00
Number of transitions Mean (SE) 0 (0.00)† 1.07 (0.03) 2.24 (0.06) 4.89 (0.12) 1.07 (0.03) 0 (0.00) †
Duration remission episodes Mean (SE) 0 (0.00) † 18.88 (0.90) 5.10 (0.28) 2.21 (0.11) 1.65 (0.31) 0 (0.00) †
Duration relapse episodes Mean (SE) 0 (0.00) † 1.27 (0.28) 4.17 (0.23) 2.84 (0.11) 18.92 (0.69) 0 (0.00) †
COMBINE Study (n=1289)
Indicators Profile 1: Continuous remission (25.3%) Profile 2: Transition to remission (19.6%) Profile 3: Few long transitions (15.9%) Profile 4: Numerous short transitions (13.2%) Profile 5: Transition to relapse (7.2%) Profile 6: Continuous relapse (18.8%)
Initial heavy drinking Proportion 0.00 0.32 0.61 0.35 0.32 1.00
Number of transitions Mean (SE) 0 (0.00)† 1.44 (0.07) 1.46 (0.07) 4.40 (0.23) 1.00 (<0.001) 0 (0.00) †
Duration remission episodes Mean (SE) 0 (0.00) † 12.34 (0.60) 4.20 (0.37) 2.57 (0.16) 3.92 (0.28) 0 (0.00) †
Duration relapse episodes Mean (SE) 0 (0.00) † 2.89 (0.17) 10.39 (0.36) 3.34 (0.18) 21.54 (0.34) 0 (0.00) †
Open in a new tab

Note.

†

remission and relapse profiles were known and constraints were placed on the model to make sure these profiles were included in the final profile solution. SE = Standard Error. Duration represents mean number of consecutive 14-day intervals for each episode (e.g., the mean relapse duration of 18.92 fourteen-day periods for profile 5 in Project MATCH corresponds to an average relapse duration of about 265 days)

COMBINE and MATCH Distal Outcome Analyses.

Next, distal outcome analyses using the manual 3-step BCH method in Mplus were conducted to examine differences in profiles on drinking outcomes and psychosocial functioning measures at 1-year (analysis n=1,111), 3-year (analysis n=676), and 7–9 years (analysis n=133) post-treatment in COMBINE (Figures 1–3). Attrition analyses indicated no significant differences between profiles in the rates of completing 1-year and 7–9 year follow-ups (p>0.05), however there were significant differences at the 3-year follow-up (χ2 (5) = 38.12, p<0.001). Those who completed the 3-year follow-up were more likely to be in profile 4 (numerous short transitions; 78.8% completion of 3-year follow-up) and less likely to be in profiles 1 (remission from heavy drinking; 46.8% completion) and 6 (continuous relapse to heavy drinking; 44.2% completion).

Figure 1.

Figure 1.

Open in a new tab

Means of outcomes in the COMBINE study at 1 year by latent profile. Note: Whiskers indicate ±SD. Effect sizes are Cohen’s d values of standardized differences between latent profiles 1–5 vs. profile 6.

Figure 3.

Figure 3.

Open in a new tab

Means of outcomes in the COMBINE Study measured at 7–9 years post-treatment by latent profile. Note Whiskers indicate ± SE. Effect sizes are Cohen’s d values of standardized differences between latent profiles 1–5 vs. profile 6.

Results from distal outcome analyses indicated profiles 1 (remission from heavy drinking) and 2 (transition to remission from heavy drinking) had the best outcomes and were not significantly different from each other on any outcomes at 1-year post-treatment. Individuals in profile 2 had significantly greater DDD and consequences at 3-years post-treatment and were significantly more likely to seek substance use disorder or mental health treatment services at 7–9 years post-treatment, as compared to profile 1. Profiles 6 (continuous relapse to heavy drinking) and 5 (transition to relapse to heavy drinking) had the worst outcomes, and profile 6 had significantly worse drinking and psychosocial outcomes at 1-year post-treatment. Profiles 6 and 5 were not significantly different from each other on nearly all outcomes at 3 years or 7–9 years post-treatment, with the only exception being that profile 6 reported significantly poorer mental health than profile 5 at 3 years post-treatment. Profiles 4 (numerous shorter transitions) and 3 (fewer longer transitions) did not differ on psychosocial functioning and DDD at any time points, but profile 3 was drinking more frequently than profile 4 at 1- and 3-year post-treatment and had greater percent heavy drinking days and more treatment seeking at 7–9 years post-treatment. Profiles 3 and 4 had better outcomes than profiles 6 and 5, as well as worse outcomes than profiles 1 and 2 on most outcomes at all timepoints. Figures 1–3 show that effects sizes for the statistically different comparisons of profiles 1–5 vs. profile 6 generally ranged in magnitude from large to small, with larger effect sizes observed for drinking-related outcomes compared to quality of life and general psychiatric distress, and less compared to more distal outcomes, respectively.

Extending prior research, we also examined the associations between profiles identified in Project MATCH and drinking outcomes and psychosocial functioning measures at 10 years post-treatment (analysis n=143; Figure 4). Attrition analyses indicated no significant differences between profiles in the rates of completing 10 year follow-ups (χ2 (5) =6.85, p=0.23). As shown in Figure 4, Profiles 1 (remission) and 2 (transition to remission) had the best outcomes and were not significantly different from each other on nearly all outcomes at 10 years post-treatment, with the only exception being significant differences on the GAATOR, with profile 2 using 12-step programs significantly more effectively than profile 1. Similarly, profiles 5 (transition to relapse to heavy drinking) and 6 (continuous relapse to heavy drinking) had the worst outcomes and only differed on the GAATOR and purpose in life, with profile 6 using 12-step recovery resources more effectively and reporting greater purpose in life. Profiles 4 (numerous shorter transitions) and 3 (fewer longer transitions) did not significantly differ on psychosocial functioning or drinking outcomes at 10 years post-treatment. Profiles 3 and 4 had better outcomes than profiles 6 and 5, and worse outcomes than profiles 1 and 2 on most outcomes. Figure 4 shows that the effect sizes of differences between profiles 1 to 5 and 6 generally were in the small to medium range, with the largest effects sizes found for differences between profiles 1 and 2, respectively, and profile 6 on drinking outcomes.

Figure 4.

Figure 4.

Open in a new tab

Means of outcomes in Project MATCH measured at 10 years post-treatment by latent profile. Note Whiskers indicate ± SE. Effect sizes are Cohen’s d values of standardized differences between latent profiles 1–5 vs. profile 6.

Discussion

We found that distinct patterns of observed transitions between remission from and relapse to heavy drinking in the first year post-treatment that were established in the MATCH outpatient sample (Maisto et al., 2018) were replicated in the COMBINE sample with a few subtle differences reflecting more heavy drinking (i.e., longer relapse periods in profile 3 and more participants in profile 6). These findings support the generalizability of the derived profiles in representing the patterns of transitions between remission from and relapse to heavy drinking that patients experience over the first year after outpatient AUD treatment. Moreover, the data showing positive longer-term outcomes associated with non-abstinence from alcohol, as well as abstinence following treatment, are consistent with research that focused on transitions to AUD remission based on DSM-IV and −5 current AUD symptoms (Dawson et al., 2009; Fan et al., 2019, respectively). The data also showed that associations between profiles and one- and three-year outcomes found in MATCH replicated in COMBINE and were extended to include additional, non-alcohol use outcomes, including recovery efforts, general health, and purpose in life in follow-ups that spanned 7–9 years in COMBINE and 10 years in MATCH.

Profiles 2, 3, and 4 (collectively 56.9% of the sample in MATCH and 48.7% of the sample in COMBINE) had one or more relapses to heavy drinking following treatment; yet all of these profiles also demonstrated the ability to transition back to remission from heavy drinking one or multiple times during the one-year period. This provides empirical evidence that a substantial portion of individuals can indeed remit from heavy drinking following one or multiple relapses to heavy drinking after treatment. Stated another way, the occurrence of one or even multiple episodes of heavy drinking following treatment does not uniformly forecast a persistent pattern of heavy drinking. Some individuals “bounce back” by having shorter or longer periods of remission from heavy drinking again and again, whereas others do not.

Moreover, individuals in profiles 2, 3, and 4 (i.e., those who “bounced back”) consistently had better long-term outcomes up to 10 years following treatment than individuals who largely remained in a relapse status following a relapse episode (profiles 5 and 6). Overall, profile 2 (transition to remission from heavy drinking), which had at least one relapse to heavy drinking, was similar to profile 1 (continuous abstinence) for most long-term outcomes across studies and timepoints, with a few notable exceptions. That is, in COMBINE profile 2 (relative to profile 1) had greater drinks per drinking day and greater drinking consequences at year 3, as greater likelihood of seeking treatment in years 7 – 9. It is also important to note that profiles 3 (few long transitions) and 4 (many short transitions), albeit having worse long-term outcomes compared to profiles 1 and 2, had better long-term outcomes than profiles 5 and 6. Altogether, these findings suggest that to the ability to repeatedly “bounce back” and regain remission from heavy drinking status, even following multiple relapses (profiles 3 and 4), is a worthwhile endeavor and can have substantial long-term benefits for patients. Furthermore, the effect sizes of differences between profiles 3 and 4 compared to profile 6 (and to profile 5, which typically showed little difference from profile 6) suggest that these long-term benefits are clinically significant.

The empirical evidence presented here on the descriptive norms of heavy drinking remission/relapse patterns, as well as the possibility and benefits of “bouncing back” from relapse episodes (even multiple episodes), may be of direct interest to individuals in recovery from AUD. Clear evidence-based knowledge about the accessibility of recovery could potentially bolster an individuals’s willingness and motivation to put in the hard work to “bounce back” from relapses during the recovery process. This evidence could also be used to help encourage patients to view any period of time without heavy drinking as potentially helpful toward their longer-term health and quality of life, even if they continue to experience episodes of heavy drinking. Furthermore, this study’s data are of particular relevance to current definitions and conceptualizations of AUD recovery as a dynamic process of change and highlight the importance of health and social functioning as well as well-being and purpose in life, given the long-term follow-up periods represented and the emphasis on functional outcomes (Cui et al., 2015; Kaskutas et al., 2014; Neale et al., 2014; White, 2007; Witkiewitz et al., in press). Therefore, the data suggest that a significant portion of individuals who have AUD and who present for AUD outpatient treatment initiate and sustain a high level of functioning not only in modification of alcohol use patterns, but also in multiple areas of psychosocial functioning, general health, and quality of life up to a decade after treatment termination. The message to communicate to individuals in treatment or in general attempting to change their alcohol use is that recovery is something that they can achieve.

This raises the questions of who achieves recovery, and how. With the replication and extension data presented in this paper, the next research step would appear to be to design theoretically-based studies that help detect hypothesized determinants of behavior change (e.g., transitions to relapse and remission) and the maintenance of behavioral changes (e.g., duration of relapse or remission episodes) in the context of AUD. Such research would be essential to predicting who is likely to “bounce back” or not from a relapse episode. Theoretical models of the maintenance of behavior change (Kwasnicka et al., 2016) are fundamental to such a research program. Kwasnicka et al.’s systematic review and synthesis of theories of health behavior change yielded several themes (p. 283) of individuals and environments that have been hypothesized to facilitate individuals’ maintainenanc of changes in health-related behavior. The themes were identified as having a sustained maintenance motive (positive outcomes of the new behavior that are valued), successfully self-regulating the health behavior in question, having ample psychological and physical resources, the new health behavior reaching “habit” status, and having strong environmental and social support for the new behavior. These themes may be readily translated into a program of quantitative reseach. Furthermore, the data of this study suggest that such a research program should emphasize intensive study of periods in the first year post-AUD treatment. For example, applying ecological momentary assessment, daily diary methods, and longitudinal data analytic methods focusing on the initiation and/or maintenance of changes in drinking would be particularly productive (Hallgren et al., 2018). Such research would begin to address what people showing different courses of remission-relapse transitions do in what contexts that affects their current use of alcohol and ultimately is associated with functioning in different life domains up to a decade later. Data that address these questions would be vital to enhancing the power of intervention methods to help individuals sustain behavior changes that they initiated in treatment by individualizing interventions according to a person’s course of change (MacKillop, 2020).

We selected the one-year follow-up in the current study given the tendency for AUD clinical trials to examine one-year outcomes as a significant follow-up point (Maisto et al., 2016), which has been based more on custom, clinical lore, and dated “relapse” (“any use”) survival curves derived originally from alcohol, heroin, and cigarette smoking treatment outcome studies (Hunt et al., 1971), rather than on more recent systematic empirical or theoretical bases. Our data provide evidence that patterns of heavy drinking remission-relapse transitions over the one-year follow-up are predictive of long term outcomes, but our data cannot address whether patterns of heavy drinking remission-relapse transitions during other periods within the first year (e.g., the first 90 days or 6 months) are as or more predictive of long-term outcomes than patterns over one-year. This question is critical clinically as well as for research. In this regard, research that evaluates transitions over intervals shorter or longer than one-year could inform efforts on the maintenance of change after treatment and could also help identify how individuals settle into patterns of transitioning between relapse and remission. Accordingly, the question of the predictive value of heavy drinking remission-relapse transitions within shorter- and longer-term periods after treatment is an important question for future research.

The current paper also has analytic limitations. We used a confirmatory approach to replicate the 6-profile solution in the COMBINE study. In this regard, as noted in the Results section and in Table 1 of the online Supplementary material, the selection of the number of profiles in a LPA solution is debateable. In this study, a balance between parsimony and fit was used, with a preference to not overextract latent profiles. The 7-profile solution fit statistics did not differ significantly from those for the 6-profile solution, and it is plausible that the former might be chosen, but we ultimately selected the 6-profile solution to avoid overextraction. It also is possible that more exploratory approaches in the COMBINE study or other clinical studies could yield different profiles of patterns of heavy drinking remission-relapse transition over time. Alternative longitudinal models, such as changepoint detection analysis (Aminikhanghahi & Cook, 2017) or regime switching models (Cabrieto et al., 2018; Chow et al., 2015), could also be used to explore latent transitions in recovery profiles. Furthermore, information on treatment received across follow-up periods was not available and could have been related to profile outcomes. The reliance on the dichotomization of drinking data into heavy and non-heavy drinking days is a limitation of this work (MacCallum et al., 2002; Pearson et al., 2016; Tueller et al., 2016). Broader definitions of recovery that incorporate continuous consumption and non-consumption outcome data (Witkiewitz et al., 2019) may be useful in future analyses. There were missing data at the long-term follow-ups and baseline characteristics, and profile membership did predict attrition. Unfortunately, the reasons for missing data can only be speculated upon. For example, perhaps individuals in profile 1 were more likely to move on from thinking about drinking and had fuller lives that prevented them from completing the follow-up, whereas individuals in profile 6 were not functioning well enough to complete the follow-up and individuals in profile 4 were still in the mix of relapse/remission cycles and thus wanted to reconnect with the COMBINE study. As noted, these are speculations, but ultimately the reasons for missing data cannot be known.. Finally, the potential impact of non-normality of indicators is a limitation and future work needs to examine the implications of indicator non-normality in repeated measures latent profile models (Sen et al., 2016).

Because this study’s data are pertinent to the recovery stage of AUD clinical course, one question for future research is whether there are differences in the AUD-relevant histories of individuals who were statistically assigned to different first-year profiles. A conceptually important question in this regard is number of prior attempts at recovery among members assigned to the different profiles, which would address the conceptualization of AUD as a chronic relapsing disease (McKay, 2009; McLellan et al., 2000). Recent data (Kelly et al., 2019) suggest that the “chronic relapsing” characterization may be applicable only to a minority of individuals with a history of AUD diagnosis (MacKillop, 2020). Related to this question is whether there are differences among individuals with different recovery attempt histories and whether these individuals engage in different activities to sustain reductions in their drinking, whether initiated by treatment or outside of treatment.

In conclusion, the data of this study replicate and extend Maisto et al.’s (2018) findings and in doing so have clear implications for the importance of viewing AUD clinical course as a process of change. Furthermore, extension of the findings to additional non-alcohol use outcomes and very-long follow-up provide information directly relevant to concepts of AUD recovery. The next steps for future research are to conduct studies aimed at advancing our understanding of determinants of alcohol use and general life functioning in the first year following treatment and how they relate to individuals’ life functioning for years afterwards.

Supplementary Material

Supplementary Table 1

Figure 2.

Figure 2.

Open in a new tab

Means of outcomes in the COMBINE Study measured at 3 years by latent profile. Note Whiskers indicate ± SE. Effect sizes are Cohen’s d values of standardized differences between latent profiles 1–5 vs. profile 6.

Public Health Significance:

This study indicates that a substantial portion of individuals with alcohol use disorder can indeed transition back to remission from heavy drinking following one or more relapses to heavy drinking during the first year after outpatient treatment. Furthermore, findings indicate that to the ability to repeatedly regain remission from heavy drinking, even following numerous relapses to heavy drinking, may confer substantial long-term benefits to individuals who have sought treatment for alcohol use disorder.

Funding:

This research was supported by a grant funded by the National Institute on Alcohol Abuse and Alcoholism R01 AA022328 (Witkiewitz, PI). The content is solely the responsibility of the authors and does not necessarily reflect the views of NIH.

Footnotes

1

Although the decision to designate an interval’s duration as two weeks is necessarily to a degree arbitrary, it has an empirical basis, first in in this team’s earlier studies of AUD relapse (Maisto et al., 1995), and in more recent research on drinking transitions (Hallgren et al., 2016).

2

Because of the potential influence of treatment received on the profiles derived, we re-ran the LPAs with treatment condition as a covariate. The results showed no change in ultimate profile solution or model selection.

References

  1. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., text rev.). Washington, DC: Author. [Google Scholar]
  2. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. [Google Scholar]
  3. Aminikhanghahi S, & Cook DJ (2017). A survey of methods for time series change point detection. Knowledge and Information Systems, 51(2), 339–367. 10.1007/s10115-016-0987-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Anton RF, O’Malley SS, Ciraulo DA, Cisler RA, Couper D, Donovan DM, … Zweben A (2006). Combined pharmacotherapies and behavioral interventions for alcohol dependence: The COMBINE study: A randomized controlled trial. JAMA, 295(17), 2003–2017. 10.1001/jama.295.17.2003 [DOI] [PubMed] [Google Scholar]
  5. Asparouhov T, & Muthén BO (2014). Auxiliary variables in mixture modeling: Using the BCH method in Mplus to estimate a distal outcome model and an arbitrary second model (No. 21; Mplus Web Notes; ). https://www.statmodel.com/download/asparouhov_muthen_2014.pdf [Google Scholar]
  6. Beck AT, Steer RA, & Carbin MG (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8(1), 77–100. 10.1016/0272-7358(88)90050-5 [DOI] [Google Scholar]
  7. Bolck A, Croon M, & Hagenaars JA (2004). Estimating latent structure models with categorical variables: One-step versus three-step estimators. Political Analysis, 12(1), 3–27. 10.1093/pan/mph001 [DOI] [Google Scholar]
  8. Bray BC, Lanza ST, & Tan X (2015). Eliminating bias in classify-analyze approaches for latent class analysis. Structural Equation Modeling, 22(1), 1–11. 10.1080/10705511.2014.935265 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brownell KD, Marlatt GA, Lichtenstein E, & Wilson GT (1986). Understanding and preventing relapse. The American Psychologist, 41(7), 765–782. 10.1037//0003-066x.41.7.765 [DOI] [PubMed] [Google Scholar]
  10. Cabrieto J, Adolf J, Tuerlinckx F, Kuppens P, & Ceulemans E (2018). Detecting long-lived autodependency changes in a multivariate system via change point detection and regime switching models. Scientific Reports, 8(1), 15637 10.1038/s41598-018-33819-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chow S-M, Witkiewitz K, Grasman RRPPP, & Maisto SA (2015). The cusp catastrophe model as cross-sectional and longitudinal mixture structural equation models. Psychological Methods, 20(1), 142–164. 10.1037/a0038962 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Crumbaugh JC, & Maholick LT (1964). An experimental study in existentialism: The psychometric approach to Frankl’s concept of noogenic neurosis. Journal of Clinical Psychology, 20(2), 200–207. [DOI] [PubMed] [Google Scholar]
  13. Cui C, Noronha A, Warren KR, Koob GF, Sinha R, Thakkar M, … Sullivan EV (2015). Brain pathways to recovery from alcohol dependence. Alcohol, 49(5), 435–452. 10.1016/j.alcohol.2015.04.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hallgren KA, Wilson AD, & Witkiewitz K (2018). Advancing analytic approaches to address key questions in mechanisms of behavior change research. Journal of Studies on Alcohol and Drugs, 79(2), 182–189. 10.15288/jsad.2018.79.182 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hunt WA, Barnett LW, & Branch LG (1971). Relapse rates in addiction programs. Journal of Clinical Psychology, 27(4), 455–456. http://www.ncbi.nlm.nih.gov/pubmed/5115648 [DOI] [PubMed] [Google Scholar]
  16. Huo T, Guo Y, Shenkman E, & Muller K (2018). Assessing the reliability of the short form 12 (SF-12) health survey in adults with mental health conditions: A report from the wellness incentive and navigation (WIN) study. Health and Quality of Life Outcomes, 16(1) 34 10.1186/s12955-018-0858-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kadden RM, Carroll KM, Donovan DM, Cooney NL, Monti PM, Abrams D, Litt MD, & Hester RK (1995). Cognitive-behavioral coping skills therapy manual. (Volume 3, Project MATCH Monograph Series). Rockville, MD: National Institute on Alcohol Abuse and Alcoholism. [Google Scholar]
  18. Kaskutas LA, Borkman TJ, Laudet A, Ritter LA, Witbrodt J, Subbaraman MS, Stunz A, & Bond J (2014). Elements that define recovery: The experiential perspective. Journal of Studies on Alcohol and Drugs, 75(6), 999–1010. 10.15288/jsad.2014.75.999 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kelly JF, Greene MC, Bergman BG, White WL, & Hoeppner BB (2019). How many recovery attempts does it take to successfully resolve an alcohol or drug problem? Estimates and correlates from a national study of recovering U.S. adults. Alcoholism: Clinical and Experimental Research, 43(7), 1533–1544. 10.1111/acer.14067 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kwasnicka D, Dombrowski SU, White M, & Sniehotta F (2016). Theoretical explanations for maintenance of behaviour change: A systematic review of behaviour theories. Health Psychology Review, 10(3), 277–296. 10.1080/17437199.2016.1151372 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Lowman C, Allen J, & Stout RL (1996). Replication and extension of Marlatt’s taxonomy of relapse precipitants: Overview of procedures and results. The Relapse Research Group. Addiction, 91(12s1), 51–72. 10.1046/j.1360-0443.91.12s1.16.x [DOI] [PubMed] [Google Scholar]
  22. MacCallum RC, Zhang S, Preacher KJ, & Rucker DD (2002). On the practice of dichotomization of quantitative variables. Psychological Methods, 7(1), 19–40. http://www.ncbi.nlm.nih.gov/pubmed/11928888 [DOI] [PubMed] [Google Scholar]
  23. MacKillop J (2020). Is addiction really a chronic relapsing disorder? Alcoholism: Clinical and Experimental Research, 44(1), 41–44. 10.1111/acer.14246 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Maisto SA, Hallgren KA, Roos CR, & Witkiewitz K (2018). Course of remission from and relapse to heavy drinking following outpatient treatment of alcohol use disorder. Drug and Alcohol Dependence, 187, 319–326. 10.1016/j.drugalcdep.2018.03.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Maisto SA, Kirouac M, & Witkiewitz K (2014). Alcohol use disorder clinical course research: Informing clinicians’ treatment planning now and in the future. Journal of Studies on Alcohol and Drugs, 75(5), 799–807. 10.15288/jsad.2014.75.799 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Maisto SA, Witkiewitz K, Moskal D, & Wilson AD (2016). Is the construct of relapse heuristic, and does it advance alcohol use disorder clinical practice? Journal of Studies on Alcohol and Drugs, 77(6), 849–858. 10.15288/jsad.2016.77.849 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. McKay JR (2009). Treating substance use disorders with adaptive continuing care. Washington, D.C.: American Psychological Association. [Google Scholar]
  28. McLellan AT, Lewis DC, O’Brien CP, & Kleber HD (2000). Drug dependence, a chronic medical illness: Implications for treatment, insurance, and outcomes evaluation. JAMA, 284(13), 1689–1695. 10.1001/jama.284.13.1689 [DOI] [PubMed] [Google Scholar]
  29. Miller WR (1996). Form 90: A structured assessment interview for drinking and related behaviors. (Vol 5, Project MATCH Monograph Series). Rockville, MD: National Institute on Alcohol Abuse and Alcoholism. [Google Scholar]
  30. Miller WR, Zweben A, DiClemente CC, & Rychtarik RG (1994). Motivational enhancement therapy manual. (Vol 2, Project MATCH Monograph Series). Rockville, MD: National Institute on Alcohol Abuse and Alcoholism. [Google Scholar]
  31. Miller WR, Tonigan JS, & Longabaugh R (1995). The Drinker Inventory of Consequences (DrInC): An instrument for assessing adverse consequences of alcohol abuse. Test manual (Volume 4, Project MATCH Monograph Series). Rockville, MD: National Institute on Alcohol Abuse and Alcoholism. [Google Scholar]
  32. Montgomery HA, Miller WR, & Tonigan SJ (1995). Does alcoholics anonymous involvement predict treatment outcome? Journal of Substance Abuse Treatment, 12(4), 241–246. 10.1016/0740-5472(95)00018-Z [DOI] [PubMed] [Google Scholar]
  33. Moos RH, & Moos BS (2005). Sixteen-year changes and stable remission among treated and untreated individuals with alcohol use disorders. Drug and Alcohol Dependence, 80(3), 337–347. 10.1016/j.drugalcdep.2005.05.001 [DOI] [PubMed] [Google Scholar]
  34. Moos RH, & Moos BS (2006). Rates and predictors of relapse after natural and treated remission from alcohol use disorders. Addiction, 101(2), 212–222. 10.1111/j.1360-0443.2006.01310.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Moos RH, & Moos BS (2007). Protective resources and long-term recovery from alcohol use disorders. Drug and Alcohol Dependence, 86(1), 46–54. 10.1016/j.drugalcdep.2006.04.015 [DOI] [PubMed] [Google Scholar]
  36. Muthén LK, & Muthén BO (2017). Mplus users guide (Version 8). Muthén & Muthén. [Google Scholar]
  37. Neale J, Finch E, Marsden J, Mitcheson L, Rose D, Strang J, … Wykes T (2014). How should we measure addiction recovery? Analysis of service provider perspectives using online Delphi groups. Drugs: Education, Prevention and Policy, 21(4), 310–323. 10.3109/09687637.2014.918089 [DOI] [Google Scholar]
  38. Nowinski J, Baker S, & Carroll KM (1995). Twelve step facilitation therapy manual. (Volume 4, Project MATCH Monograph Series). Rockville, MD: National Institute on Alcohol Abuse and Alcoholism. [Google Scholar]
  39. Nylund-Gibson K, Grimm RP, & Masyn KE (2019). Prediction from latent classes: A demonstration of different approaches to include distal outcomes in mixture models. Structural Equation Modeling: A Multidisciplinary Journal, 26(6), 967–985. 10.1080/10705511.2019.1590146 [DOI] [Google Scholar]
  40. Nylund KL, Asparouhov T, & Muthén BO (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535–569. 10.1080/1070551070157396 [DOI] [Google Scholar]
  41. Pagano ME, White WL, Kelly JF, Stout RL, & Tonigan JS (2013). The 10-year course of Alcoholics Anonymous participation and long-term outcomes: A follow-up study of outpatient subjects in Project MATCH. Substance Abuse, 34(1), 51–59. 10.1080/08897077.2012.691450 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Pearson MR, Kirouac M, & Witkiewitz K (2016). Questionning the validity of the 4+/5+ binge or heavy drinking criterion in college and clinical populations. Addiction, 111(10), 1720–1726. 10.1111/add.13210 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Project MATCH Research Group. (1997). Matching Alcoholism Treatments to Client Heterogeneity: Project MATCH posttreatment drinking outcomes. Journal of Studies on Alcohol, 58(1), 7–29. 10.15288/jsa.1997.58.7 [DOI] [PubMed] [Google Scholar]
  44. Sen S, Cohen AS, & Kim SH (2016). The impact of non-normality on extraction of spurious latent classes in mixture IRT models. Applied Psychological Measurement, 40(2), 98–113. 10.1177/0146621615605080 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Skinner HA, & Horn JL (1984). Alcohol Dependence Scale (ADS): User’s Guide. Toronto, Canada: Addiction Research Foundation. [Google Scholar]
  46. Spielberger CD (1988). Manual for the State-Trait Anger Expression Scale. Odessa, FL: Psychological Assessment Resources. [Google Scholar]
  47. Tueller SJ, Deboeck PR, & Van Dorn RA (2016). Getting less of what you want: Reductions in statistical power and increased bias when categorizing medication adherence data. Journal of Behavioral Medicine, 39(6), 969–980. 10.1007/s10865-016-9727-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Ware JE, Kosinski M, & Keller SD (1996). A 12-Item Short-Form Health Survey: Construction of scales and preliminary tests of reliability and validity. Medical Care, 34(3), 220–233. 10.1097/00005650-199603000-00003 [DOI] [PubMed] [Google Scholar]
  49. White WL (2007). Addiction recovery: Its definition and conceptual boundaries. Journal of Substance Abuse Treatment, 33(3), 229–241. 10.1016/j.jsat.2007.04.015 [DOI] [PubMed] [Google Scholar]
  50. WHOQOL Group. (1998). Development of the World Health Organization WHOQOL-BREF Quality of Life Assessment. Psychological Medicine, 28(3), 551–558. 10.1017/S0033291798006667 [DOI] [PubMed] [Google Scholar]
  51. Witkiewitz K, Finney JW, Harris AHS, Kivlahan DR, & Kranzler HR (2015a). Guidelines for the reporting of treatment trials for alcohol use disorders. Alcoholism: Clinical & Experimental Research, 39(9), 1571–1581. 10.1111/acer.12797 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Witkiewitz K, Finney JW, Harris AHS, Kivlahan DR, & Kranzler HR (2015b). Recommendations for the design and analysis of treatment trials for alcohol use disorders. Alcoholism: Clinical and Experimental Research, 39(9), 1557–1570. 10.1111/acer.12800 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Witkiewitz K, Montes KS, Schwebel FJ, & Tucker JA (n.d.). What is recovery? A narrative review of definitions of recovery from alcohol use disorder. Alcohol Research: Current Reviews. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Witkiewitz K, Wilson AD, Pearson MR, Montes KS, Kirouac M, Roos CR, Hallgren KA, & Maisto SA (2019). Profiles of recovery from alcohol use disorder at three years following treatment: Can the definition of recovery be extended to include high functioning heavy drinkers? Addiction, 114(1), 69–80. 10.1111/add.14403 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Zarkin GA, Bray JW, Aldridge A, Mills M, Cisler RA, Couper D, … OʼMalley S (2010). The effect of alcohol treatment on social costs of alcohol dependence. Medical Care, 48(5), 396–401. 10.1097/MLR.0b013e3181d68859 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Zarkin GA, Bray JW, Aldridge A, Mitra D, Mills MJ, Couper DJ, & Cisler RA (2008). Cost and cost-effectiveness of the COMBINE study in alcohol-dependent patients. Archives of General Psychiatry, 65(10), 1214–1221. 10.1001/archpsyc.65.10.1214 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Table 1
NIHMS1664580-supplement-Supplementary_Table_1.docx (14KB, docx)

RESOURCES