This quasi-experimental study of the MAAEZ intervention in real-world treatment programs found significantly higher rates of past 30-day alcohol abstinence, drug abstinence, and abstinence from both alcohol and drugs for respondents in the ON condition at the 12-month follow-up interview compared to those in the OFF. The magnitude of the MAAEZ effect was similar to those reported in other TSF studies such as Project MATCH (Tonigan et al., 2003
) and the intensive referral study (Timko & Debenedetti, 2007
; Timko et al., 2006
): an abstinence rate about 8% to 10% higher than that of the comparison group(s). Both intent-to-treat and as-treated multivariate analyses found a MAAEZ effect for alcohol abstinence and for drug abstinence; the as-treated analysis found a significant dose effect for total abstinence as well. The study was conducted in outpatient programs, long-term residential programs, and a short-term residential program, where existing recovering counselors delivered MAAEZ with high fidelity to the manual. Findings were consistent whether or not we controlled for program type and length of stay. The study programs represent hybrid 12-step-oriented programs for treating both alcohol and drug dependent clients, and participants were heterogeneous in terms of demographics, diagnosis, and prior service utilization. Given the robust 12-month bivariate results and the consistent multivariate results obtained across techniques and outcomes, we conclude that MAAEZ represents an evidence-based TSF intervention, widely applicable to the chemical dependency treatment field.
The especially high abstinence rate (92%) among the 59 participants who attended all 6 MAAEZ sessions may point to the power of peer helping that MAAEZ participants found when returning to the Introductory
session to share their MAAEZ and AA experience with MAAEZ newcomers. It also could be an artifact of higher motivation among those attending more sessions, although our multivariate analyses controlling for length of stay also found an effect for higher MAAEZ dose. The result is consistent with earlier work we have done on helping, which found that peer helping during treatment was associated with more AA attendance following treatment, which in turn led to better outcomes (Zemore & Kaskutas, 2008
; Zemore et al., 2004
). Given the emphasis on professional counseling, peer helping may be underemphasized in treatment programs, but could be an effective tool that is easily implemented.
Guided by findings of superior effects among specific subgroups for prior TSF interventions, we also conducted extensive disaggregated analyses. Our findings are consistent with positive TSF effects for one subgroup considered in the literature, those with high psychiatric severity. We found a strong, consistent MAAEZ effect among those with ASI psychiatric problem severity at or above the median in our sample. This is similar to Brown et al., who found that TSF was particularly effective for those with high psychological distress (Brown et al., 2002
). However, other TSF studies have had different results with respect to psychiatric comorbidity. Those with psychopathology did equally well in TSF and CBT in Project MATCH, but depressed clients did better in CBT than in TSF in Maude-Griffin’s study (Maude-Griffin et al., 1998
). Many have expressed concern that AA may be inappropriate for those with psychiatric problems (American Psychiatric Association, 1995
; Noordsy, Schwab, Fox, & Drake, 1996
), but empirical evidence of AA’s effectiveness has begun to paint a different picture (Morgenstern et al., 2003
; Morgenstern, Kahler, & Epstein, 1998
; Ouimette, Gima, Moos, & Finney, 1999
). Given the high level of concurrent mental health disorders among treatment-seeking clients, MAAEZ is especially promising for those at the high end of the severity spectrum.
Our results for prior AA exposure also diverge from prior work on TSF effectiveness. For example, TSF results in Project MATCH and the intensive referral study favored those with less prior AA meeting exposure, while the opposite was the case with MAAEZ. Furthermore, we found that MAAEZ was more effective among those with prior treatment episodes. MAAEZ’s emphasis on connecting with the AA fellowship and culture may have given these recidivistic clients a new way of viewing AA and AA members that had been unfamiliar to them in past experiences with AA. Given the high AA drop-out rates (i.e., about half of newcomers drop out in less than a month; Alcoholics Anonymous, 1990
) and the recidivistic nature of treatment (72% among those recruited in long-term residential programs in our study alone), interventions that tangibly help clients to try AA again, with an open mind and greater involvement with AA members, are needed. Based on anecdotal evidence and our own observations at AA meetings and among treatment clients, we believe there are many individuals who attend AA but tend to sit around the edges, arriving late and leaving early and never connecting much with the people in the program. We believe MAAEZ meets the needs of these people especially. Book and Randall (2002)
have noted that social anxiety pervades among many alcoholics, so it makes sense that such individuals may not know how to introduce themselves to people at AA. Timko’s intensive referral intervention connected clients with an AA member who met them at a meeting, which may have been just what a neophyte needs to spark a relationship with AA. MAAEZ results suggest that those with more prior experience in AA may need more than that. Similarly, in Project MATCH, clients learned about acceptance and surrender, concepts probably helpful to AA newcomers but somewhat “old hat” to those with lots of AA exposure. This interpretation is supported by Kahler’s finding of a positive effect for their very brief TSF intervention only among those with minimal prior AA/NA experience (Kahler et al., 2004
). Since different types of TSF appear necessary based on past AA experience, providers may want to consider matching different TSF approaches based on prior AA exposure.
Our results also differed from prior research in terms of religiosity. Maude-Griffin found that those with strong religious beliefs did better in TSF, while we found a greater MAAEZ effect for those who were atheist, agnostic or unsure of their religious beliefs. Maude-Griffin studied African American clients, for whom the church and religiosity are powerful sources of social and emotional support. In contrast, many of our participants were secular, and our MAAEZ spirituality session did focus on introducing clients to a new, more neutral and accepting way of experiencing the talk about god and spirituality that they inevitably encountered in AA.
We also found stronger MAAEZ effects among demographic subgroups not previously identified in prior TSF studies. These included males, non-married individuals, and those without high school diplomas. The social encouragement from MAAEZ homework may have been especially helpful for those living alone or without a partner. Further, the practical homework assignments may have appealed most to those with less formal education who may become frustrated by the didactic nature of some substance abuse treatment (Kaskutas, Ammon, Witbrodt et al., 2005
Other subgroups with strong MAAEZ effects were those with drug problems at or above median severity, and those whose alcohol problems were below median severity; this is a puzzling pair of findings. Perhaps neither group had seriously considered AA, thinking it was only for alcoholics, until they heard in MAAEZ that they could go to AA, and would be welcomed. MAAEZ may have also made them feel more comfortable in AA. Our study included those with alcohol and/or drug problems, so there may have been considerable overlap of individuals within the high drug severity and low alcohol severity groupings.
Last, we found a greater MAAEZ effect for those without heavy alcohol and drug users in their social network at baseline. At its 3-year follow-up, Project MATCH found a greater TSF effect among outpatients with high network support for drinking at study entry. It may be that having fewer heavy drinkers in one’s social network takes precedence during the first year of abstinence, while having skills to cope with heavy drinkers is more important 3 years after treatment. Perhaps Project MATCH TSF inoculated participants from negative pressure from heavy drinkers, whereas MAAEZ reinforced the positive influence of social networks without any heavy drinkers or users. We plan to explore social network variables (including support for drinking, support for quitting, support from 12-step members) in forthcoming mediational analyses of MAAEZ. MAAEZ is grounded in the Theory of Planned Behavior (Ajzen, 1985
), and aims to overcome resistance to 12-step groups by changing participants’ attitudes, subjective norms and perceived control regarding 12-step involvement. Thus we will study not only changes in social networks, but also how clients viewed what would happen (including what their friends would think) if they became involved in AA/NA/CA. We also will look in-depth at how MAAEZ participation was associated with AA involvement, a key theoretical mediator of MAAEZ’s effects. Evidence of reasonable mechanisms of action, that are consistent with our theoretical model, would lend further confidence in the effectiveness of MAAEZ.
Although the study had a strong response rate at the 12 month follow-up interview (76%), an attrition bias was present that limits the generalizability of these results. Individuals with a lower socioeconomic status and who only had a drug diagnosis were under-represented at follow-up. However, since the MAAEZ effect was stronger among those with severe drug problems, this implies that our results actually may underestimate MAAEZ effectiveness. In addition, our quasi-experimental design had less power than a traditional randomized trial. However, randomization would have introduced contamination and thus reduced the internal validity of such a trial. Our historical content analysis, which tracked potential events that might have impacted the effectiveness of 12-step facilitation, noted both increases and decreases in substance abuse treatment funding as well as high-profile entrants into treatment during the recruitment period. Still, our empirical comparison of the two OFF conditions found no difference in outcomes that might suggest a historical effect, so we feel confident that MAAEZ effectiveness is not an artifact of the study design.
We also conducted multiple subgroup analyses, introducing the potential for Type I error. The consistency of results between the intent-to-treat and as-treated analyses mitigates this concern somewhat, and suggests some robustness of effect within particular subgroups. Another limitation was our inability to widely validate self-reported drug use in the sample as had been planned, and the discordance between drug tests and self-reported drug use among those few who did provide urine samples and tested positive for current drug use. The rates of past 30-day abstinence from alcohol and drugs in our study are about 10%-15% higher than those reported in other studies (see, for example, (Weisner, Mertens, Parthsarathy, & Moore, 2001
; Witbrodt et al., 2007
). Still, we have no reason to believe that false positives would have been more common in the ON condition, so this issue should not detract from the findings regarding MAAEZ effectiveness. One final limitation is the small number of ON condition clients who received a full MAAEZ dose (barely 20%). This reflects the reality of service delivery in real-world treatment programs, in which clients cannot be expected to complete treatment. Thus, this health services trial of MAAEZ has strong external validity.
Our analyses did not account for the potential impact of the group dynamic on outcome (Kenny, Mannetti, Pierro, Livi, & Kashy, 2002
). In group-administered treatment, participants interact with one another and can influence one another’s behavior, causing interdependence of observations. Statistical models can address this problem by considering the intra-class correlation (ICC) among those in the same group to judge significance (Baldwin, Murray, & Shadish, 2005
; Kenny & Judd, 1986
; Morgan-Lopez & Fals-Stewart, 2006
). Failure to account for this group effect can increase the likelihood of finding a statistically significant difference between treatment conditions that appears to be due to condition, but is in fact caused by the group effect. Even a small group effect (ICC=.05) can increase type 1 error by 20% (Varnell, Murray, Hannan, & Baker, 2001
). The bias associated with the group dynamic increases as group size, number of sessions, and group stability increase (Baldwin et al., 2005
; Kenny & Judd, 1986
; Kenny et al., 2002
We considered whether we needed to correct for ICC in our study. Group sizes were modest, with about 6 at the introductory sessions and 12 at the core sessions, and the number of sessions was small (6). Also, group stability was low: 25%-75% turn-over at each core session attended, and 50% turn-over at each introduction session, with half newcomers and half attending as graduates. Given the limited size of the groups, the small number of sessions, and the high turn-over arising from the intervention’s design, it is not clear that we should expect a group effect at all from MAAEZ. Furthermore, power for detecting significant condition differences within the latent classes would be limited were we to correct for ICC, such that we would be unable to determine definitively whether results discordant from those presented here were due to ICC or power. Because ON condition clients returned to usual care after the MAAEZ intervention, still another group dynamic was encountered by study participants. For these reasons (some of which flow from the structure of MAAEZ, others from the study design and available sample), we decided not to implement a statistical adjustment for group membership in our study.
MAAEZ participation led to higher abstinence rates in the sample overall and in important subgroups including unmarried individuals, those with more severe psychiatric problems, individuals who were atheists, agnostics, or unsure of their religious beliefs, clients who had been to treatment before, and those with more prior AA exposure. MAAEZ is brief and easily implemented in existing residential and outpatient treatment programs by simply laminating the MAAEZ session outlines included in the manual. The manual is available free of cost from the senior author.