If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Kaiser Permanente Division of Research, 2000 Broadway, 3rd floor, Oakland, CA 94612, USADepartment of Psychiatry, University of California, San Francisco, 401 Parnassus Ave, Box F-0984, San Francisco, CA 94143, USA
This study explored causal relationships between post-treatment 12-step attendance and abstinence at multiple data waves and examined indirect paths leading from treatment initiation to abstinence 9-years later. Adults (N = 1945) seeking help for alcohol or drug use disorders from integrated healthcare organization outpatient treatment programs were followed at 1-, 5-, 7- and 9-years. Path modeling with cross-lagged partial regression coefficients was used to test causal relationships. Cross-lagged paths indicated greater 12-step attendance during years 1 and 5 and were casually related to past-30-day abstinence at years 5 and 7 respectfully, suggesting 12-step attendance leads to abstinence (but not vice versa) well into the post-treatment period. Some gender differences were found in these relationships. Three significant time-lagged, indirect paths emerged linking treatment duration to year-9 abstinence. Conclusions are discussed in the context of other studies using longitudinal designs. For outpatient clients, results reinforce the value of lengthier treatment duration and 12-step attendance in year 1.
) that centered on the effectiveness of Alcoholics Anonymous (AA) has been challenged in recent years by a substantial empirical literature suggesting that participation in AA (
in: Recent developments in alcoholism: Research on Alcoholics Anonymous and spirituality in addiction recovery. Vol. 18. Springer,
New York2008: 357-372
Attendance at Narcotics Anonymous and Alcoholics Anonymous meetings, frequency of attendance and substance use outcomes after residential treatment for drug dependence: A 5 year follow-up study.
). Research on 12-step programs in the past decade has become increasingly sophisticated both in study design and analytic approach. Consistent and robust findings indicating a positive correlation between AA participation and improved outcomes have been further advanced by findings from a handful of longitudinal studies using repeated and extended follow-ups, including some that have used time-lagged designs to test causal relationships between AA exposure and outcomes (
A longitudinal model of intake symptomatology, AA participation, and outcome: Retrospective study of the Project MATCH outpatient and aftercare samples.
Alcoholics Anonymous involvement and positive alcohol-related outcomes: Cause, consequence, or just a correlate? A prospective 2-year study of 2,319 alcohol-dependent men.
Journal of Consulting and Clinical Psychology.2003; 71: 302-308
We might ask what the role of treatment is in the relationship between AA participation and improved outcomes as many help-seekers transition between treatment and 12-step programs before obtaining sustained remission (
). Studies of treatment and general population samples also suggest that specialty treatment and AA and other 12-step groups are closely linked when viewed concurrently (
Epidemiology of Alcoholics Anonymous participation.
in: Recent developments in alcoholism: Research on Alcoholics Anonymous and spiritual aspects in addiction recovery. Vol. 18. Springer,
New York2008: 261-282
). While little data are available to substantiate whether AA itself is more effective than specialty treatment in general, the literature does suggest that treatment in combination with 12-step participation, especially treatment based on 12-step principles and behaviors, may be more successful in effecting abstinence over time (
). This paper explores the relationship between 12-step meeting attendance and abstinence using longitudinal data collected from help-seeking individuals with alcohol and other drug use disorders who were recruited from a private, integrated health system at treatment entry and who were interviewed at five data points across the span of nearly a decade. Integrated and other managed care organizations have become major providers for the private and public sectors, yet, only a few longitudinal studies of 12-step participation have focused on these populations (
Twelve-step affiliation and three-year substance use outcomes among adolescents: Social support and religious service attendance as potential mediators.
). Our analytic strategy is theoretically guided by the clinical and longitudinal treatment outcome research, including treatment careers and the natural course of treated populations (
Our analysis builds mainly on a set of longitudinal studies. Three of these utilized cross-lagged analytic techniques to evaluate causal paths between AA participation and better outcomes. Under this analytical method reciprocal causation is explicitly modeled (
Alcoholics Anonymous involvement and positive alcohol-related outcomes: Cause, consequence, or just a correlate? A prospective 2-year study of 2,319 alcohol-dependent men.
Journal of Consulting and Clinical Psychology.2003; 71: 302-308
was the first to apply a cross-lagged design to evaluate whether AA involvement was a “cause, consequence or merely a correlation” of better outcomes. Results indicated that higher levels of AA involvement in year 1 predicted better alcohol-related outcomes in year 2, but alcohol-related outcomes in year 1 did not predict AA involvement in year 2. These results were obtained even after controlling for baseline levels of AA involvement and problem severity. Treatment involvement was not considered in the model; however, the causal relationship between AA involvement in year 1 and the drinking outcomes in year 2 was not altered when baseline motivation was controlled. In a similar study,
analyzed alcohol-dependent, outpatient clients (n = 169) randomized to one of three conditions: a directive approach to facilitating AA, a motivational enhancement approach to facilitating AA, or treatment as usual (no emphasis on AA). Cross-lagged panel models showed that AA involvement during treatment predicted alcohol abstinence at 4–6 months after treatment end, and AA involvement at 4–6 months after treatment end predicted alcohol abstinence at 10–12 months after treatment end. Similar to
Alcoholics Anonymous involvement and positive alcohol-related outcomes: Cause, consequence, or just a correlate? A prospective 2-year study of 2,319 alcohol-dependent men.
Journal of Consulting and Clinical Psychology.2003; 71: 302-308
found that AA attendance at 3-, 6-, 9- and 12-months predicted increases in alcohol abstinence and reduction in drinking problems at 6-, 9-, 12- and 15-months respectively for the outpatient subsample (n = 952). Again casual effects in the reverse direction were unsupported. Results for the aftercare subsample (n = 744) were not as clear, but they also suggested that AA attendance leads to better outcomes. The current study builds on these cross-lagged studies by extending out to 9 years with a privately insured sample seeking services from a program that incorporated 12-step ideology into some group sessions.
Our analysis is also informed by findings from three extended, multi-wave studies. One such study followed initially untreated individuals (n = 628) with alcohol use disorders for 16 years and examined the sequence and duration of treatment and 12-step participation in relation to alcohol-related outcomes at three follow-ups. Better results were found at both 1 and 8 years for individuals who initiated treatment and 12-step participation in year 1; a longer duration of treatment in year 1 independently related to a higher likelihood of improved outcomes at these two follow-ups (
Whereas this 16-year study analyzed data between two distinct time points, our current analysis is also informed by longitudinal studies that simultaneously considered multiple data waves in a single statistical model (
). These generally suggested that patterns of 12-step attendance vary across individuals and that ongoing participation is associated with better outcomes. Using latent class trajectories analysis (LCA) with the current integrated care treatment sample of alcohol and drug misusers, authors found that individuals reporting high steady 12-step attendance over 9 years and those reporting initial high levels of attendance followed by low but continued attendance reported the highest stable patterns of past-30 alcohol and drug abstinence over time; in contrast, those in a class with no (or very little) 12-step attendance over time reported the lowest pattern of abstinence over time (
). Longer index treatment duration was associated with membership in classes with the highest initial 12-step attendance patterns, and female gender and high problem severity correlated with 12-step attendance at all waves.
Drawing from these longitudinal studies, we use a path modeling strategy that incorporates cross-lagged and autoregressive techniques to assess causal linkages between treatment, 12-step attendance and abstinence over time. In this analytical approach each variable in the hypothesized path is regressed on all variables that precede it in time thus making full use of the multi-wave longitudinal data. While prior multi-wave longitudinal studies have examined these issues, they have not specifically explored the temporal and cumulative relationships between post-treatment 12-step attendance and treatment readmissions and abstinence status over time while considering the influence of index treatment duration as we do here. It is important to include the role of treatment because of its potential to confound the relationship between 12-step attendance and abstinence. Lastly, several papers examining the sample included in the current study support our conceptual model, but none looked at the cumulative relationships between treatment, 12-step attendance and abstinence over time (
Following on the hypothesized directional relationships described in the three studies that used cross-lagged analyses, this method allowed us to explore (1) whether prior 12-step attendance is in the casual path leading to subsequent abstinence or, conversely, whether prior abstinence is the causal path leading to subsequent 12-step attendance for each of three time-lagged periods; or (2) whether these hypothetical relationships change over time (if they are cyclical). Similar to the
Alcoholics Anonymous involvement and positive alcohol-related outcomes: Cause, consequence, or just a correlate? A prospective 2-year study of 2,319 alcohol-dependent men.
Journal of Consulting and Clinical Psychology.2003; 71: 302-308
study, we controlled for baseline problem severity and also included treatment duration in the model. The model additionally allowed us to explore not only the directional direct effects of formal treatment on 12-step attendance and abstinence, but also the time-lagged indirect paths leading from the point treatment initiation to abstinence 9 years later.
2. Materials and methods
2.1 Study site
Study individuals (N = 1951) were patients from two randomized controlled trials that were conducted at the Kaiser Permanente Chemical Dependency Recovery Program (CDRP) in Sacramento, CA (
). Kaiser Permanente is a private group-model integrated health care organization covering 40% of the Sacramento catchment area population. It provides substance use and psychiatric services internally. Most members are insured through their own or a family member's employer. This CDRP site was chosen for the two studies because it serves a heterogeneous client population. Recruitment periods occurred between the years 1994 and 1998. We combined the two samples to obtain better statistical power to conduct our analyses. Other papers have reported on these combined samples (
). The CDRP provided traditional outpatient and day treatment programs. Both programs were group- and abstinence- based and offered supportive therapy, education, relapse prevention and family therapy. Individual counseling was available as needed. Content, structure and staffing were the same; only the intensity differed (day treatment included four times the amount of each service). Usual care in both programs was defined as 8 weeks of rehabilitation with 10 months of aftercare available (
). Patients (in both programs and both studies) were required to attend 12-step meetings (or a similar substitute) in the community each week and to attend an onsite meeting.
2.2 Procedure
Both studies used the same procedures, recruitment methods and follow-up study design. Patients were referred to treatment through several sources including medical providers, employers and employee assistance programs; or they were self-referred. Patients 18 and older who requested treatment and who met standard treatment admission criteria were eligible for study participation. Patients with dementia, mental retardation or active psychosis were not eligible for the programs or either study (
Research staff independent of the CDRP introduced the study and gained informed consent from patients deemed medically ready and not in need of detoxification by medical staff who conducted the initial intake assessment. Those refusing randomization or unable to be randomized, but agreeing to participate in other aspects of the studies, were also recruited and received “treatment assignment as usual.” Most study participants were randomized to treatment intensity or chose one option based on factors such as their employment schedule (
). Research staff conducted follow-up interviews by telephone at 1, 5, 7 and 9 years (response rates of 86, 78, 72, and 69%, respectively). Over 9 years, 8% died, 8% refused the study and 14% were not located. Refer to (
) for further details. Annual Institutional Review Board approval was obtained from the Kaiser Research Foundation Institute and the University of California, San Francisco.
2.3 Variables in the path model
2.3.1 Formal treatment duration
Data extracted from the health plan's automated databases were used to obtain information on CDRP treatment utilization. Treatment duration for the index episode was measured as the number of days between intake and the last treatment visit (
). Total readmissions to a CDRP and self-reported treatment admissions to programs outside CDRP were dichotomized (0,1). A similar approach was used to study readmissions in prior work (
Individuals self-reported the total number of 12-step meetings they had attended over the prior 6 months at the 1-year interview and over the prior 12-months at subsequent interviews. Twelve-step meetings were defined as attendance at groups like Alcoholics Anonymous, Narcotics Anonymous, and Cocaine Anonymous. These variables were capped at 365 to adjust for a few outliers (< 10% at any interview) and an average percent-time-in-attendance was calculated by dividing the number of meetings attended by total days in the follow-up period. Measures were transformed at analysis using a natural log transformation to normalize the distribution (
Individuals self-reported whether they had been abstinent from alcohol and other drugs (i.e., illicit and non-prescribed drug use) in the 30 days prior to each follow-up. This measure is consistent with the abstinence-based philosophy of 12-step groups and the intended goal of CDRP. Random urine screens for detection of alcohol and drug use and breath analysis for alcohol use were conducted at all follow-ups to validate self-reported data. Drug test results found good validity for self-report data (
Demographic characteristics collected at baseline (see below) were used to describe the sample and to test for attrition bias at every follow-up. Baseline past-30 day problem severity was measured using Addiction Severity Index (ASI) composite scores for the alcohol, drug and psychiatric domains. ASI values range from 0 (no problems) to 1 (high severity). The ASI has been shown a reliable and valid instrument (
Review of temporal effects and outcome predictors in substance abuse treatment studies with long-term follow-ups: Preliminary results and methodological issues.
) and employed to simultaneously estimate direct and indirect (mediating) effects. In addition to constructing paths guided by the reviewed literature, these were also informed by bivariate results from Pearson product–moment correlation tests. We used the theta parameterization because at least one dependent variable was categorical (
) when the model estimated contains one or more categorical variables (e.g. the dichotomous abstinence variables). Mplus retains all cases with sufficient data to estimate a path model (minimally 2 data points) under the assumption that data are missing at random (
). For indirect paths, Mplus provides both estimates and appropriate significance tests for situations when either the mediator or the outcome variable is considered to be continuous or categorical (
The baseline sample (N = 1951; results not displayed) was composed of more males (64%) than females (36%). The majority were Caucasian (74%) or African-American (11%). At intake the mean age was 37 years (SD = 10.8), over half were employed (60%), just under half were in a marital-like relationship (45%) and the majority had a high school or greater education (48% had a high school diploma/GED and 37% had advanced educations). Baseline interview questions adapted from the Diagnostic Interview Schedule for Psychoactive Substance Dependence provided a diagnosis of alcohol and drug dependence and abuse for each of nine substances (
). Forty percent were diagnosed as alcohol dependent, 29% drug dependent, and 19% both alcohol and drug dependent; 12% were given a substance misuse (unspecified) diagnosis. Baseline, average ASI scores were 0.42 (SD = 0.44) for alcohol, 0.13 (SD = 0.43) for drug and 0.40 (SD = 0.40) for psychiatric. In a 2000 study of KP members in the general membership and members from treatment programs the mean values (and standard deviations) for alcohol, drug and psychiatric ASI composite scores were 0.38 (0.32), 0.11 (0.12), and 0.37 (0.24) respectively for the treatment sample and 0.11 (0.07), 0.01 (0.03), and 0.03 (0.10) respectively for the general membership (
). Severity ratings across domains are not intended to be compared.
Over half the sample (58%) were interviewed at all four follow-ups (mean follow-ups = 3). Individuals with missing follow-up data were more likely to be men and to have a shorter average index treatment duration (at every follow-up); and they reported less education, lower income (two follow-ups) and less employment (three follow-ups). ASI problem severities did not differentiate attrition status. Because our statistical model assumed that data were missing at random and there is no test for establishing whether data are missing at random (
), we added those variables that correlated with attrition (five total) to our path model to control for their effects on AA attendance and abstinence outcomes at every follow-up.
3.2 Descriptive statistics for path model variables
The median index treatment duration was 21 days (Table 1). Excluding consented individuals who did not return for the rehabilitation phase of treatment (16%), the median duration rose to 27 days (not shown). Less than one-fifth (16%) sought additional treatment between years 2 and 5. Just over half (56%) of those interviewed at year 1 reported attending 12-step meetings. Attendance dropped after that, with nearly two-fifths (38%) of those interviewed at the 5-year follow-up reporting any attendance in the prior year and slightly less (35%) at the next two follow-ups. The average percent-time-in-attendance among attendees was fairly stable, going from 29% at year 1 (about nine meetings a month on average) and reducing slightly at subsequent interviews (22% at year 9; about six to seven meetings a month on average). Like the 12-step attendance measures, reported past-30-day alcohol and drug abstinence was highest at the 1-year follow-up (60%) and dropped after that (55% at year 9).
Table 1Treatment involvement, 12-step attendance and abstinence at 1-, 5-, 7- and 9-year follow-ups.
Fig. 1 displays standardized partial regression coefficients for significant (p < .05) direct effects (solid lines) between the treatment involvement (index treatment duration and dichotomized readmissions), 12-step attendance (percent time in attendance) and the past-30-day alcohol and drug abstinence variables. Dashed lines indicate paths that were tested in the simultaneous model, but were not significant. Direct effects (residual weights that adjust for the influences of all prior variables in the casual path) are those with arrows from one variable to another (e.g. index treatment duration →12-step Y1). The model yielded good overall model fit (χ2 = 140.4 (p < .001); CFI = 0.96; RMSEA= .040). Contemporaneous 12-step attendance and abstinence variables were allowed to be correlated over time as illustrated by bi-directional arrows in Fig. 1. ASI alcohol, drug and psychiatric scores (e.g. ASI alcohol severity → index treatment) and covariates associated with attrition (e.g. gender → abstinenceyr1; gender → 12-step attendance yr1) in the path model are not displayed in the figure.
Fig. 1Standardized partial beta coefficients for a path model assessing cross-lagged relationships between index atreatment duration, b12-step participation (percent time in attendance), cpast 30-day alcohol and drug abstinence (0,1) and dreadmissions to treatment (0,1) over follow-up periods, controlling for baseline severity and covariates associated with attrition. Statistics for significant direct effects are shown as solid lines: dashed lines indicate non-significant paths in the model.
3.4 Cross-lagged effects between 12-step attendance and abstinence
Controlling for the influence of prior effects, two significant casual associations were detected for the cross-lagged relationships tested (Fig. 1): post-treatment 12-step attendance at year 1 was causally related to past-30-day abstinence at year 5 (12-step Yr 1→abstain Yr 5;β = 0.17; p < .001), and 12-step attendance at year 5 was causally related to past-30-day abstinence at year 7 (12-step Yr 5→abstain Yr 7;β = 0.07; p = .039). These findings support the directional relationships found in the three studies that also employed cross-lagged designs (but between shorter and less extended follow-ups).
3.5 Treatment effects
Strong direct effects were found for index-treatment duration on both 12-step attendance (β = 0.45; p < .001) and abstinence (β = 0.44; p < .001) at year 1. As well, treatment readmission was related to 12-step attendance at all follow-ups (β's = 0.27, 0.17, and 0.20; p < .001) and abstinence at year 5 (β = 0.34, p = .001). Significant treatment relationships support our decision to include them as possible confounders in the path model. Prior 12-step attendance had a strong positive relationship with subsequent attendance at each subsequent follow-up (β's = 0.47, 0.67 and 0.66; p < .001) and, likewise, prior abstinence was strongly related to subsequent abstinence (β's = 0.50, 0.77 and 0.84; p < .001). Effect sizes for 12-step attendance and abstinence increased after year 5 (relationships became stronger with time), but more so for abstinence.
3.6 Indirect lagged paths related to abstinence at year 9
Tests of significance for indirect time-lagged effects conducted in the model confirmed that a longer index treatment duration was causally related to abstinence at year 9 via three overlapping paths (Table 2): (1) past-30 day abstinence at one follow-up related to past-30 day abstinence at a subsequent follow-up; (2) 12-step attendance in year 1 related to abstinence at year 5, which then followed along the abstinence path; (3) and similarly 12-step attendance in year 1 related to attendance in year 5, which too followed along the abstinence path.
Table 2Indirect time-lagged paths from baseline/index treatment to 9-year abstention.
StdYX estimate
Sign.
abIndex treatment →abstain Yr 1→abstain Yr 5→abstain Yr 7→ abstain Yr 9
.144
< .001
abIndex treatment→12-step Yr 1→abstain Yr 5→abstain Yr 7→abstain Yr 9
.048
< .001
bIndex treatment→12-step Yr 1→12step Yr 5→abstain Yr 7→abstain Yr 9
.013
.054
Paths significant for awomen or bmen in disaggreagated models.
For interpretive purposes, auxiliary analyses were conducted to test whether individuals readmitted to treatment had attended 12-step meetings in the prior periods at rates equivalent to those who did not seek out additional treatment. In every case, those who returned for additional treatment reported attending significantly fewer 12-step meetings in the period preceding their treatment reentry. Readmissions were associated with 12-step attendance at all follow-ups.
Also, auxiliary path models were estimated stratified on gender to see if results of the mixed-gender model would replicate for both genders. Fit statistics for these gender-specific models were similar to those in the mixed-gender model (CFI > .95; RMSEA < .05). Model relationships and estimates for men looked very much like the mixed-gender model, that is, cross-lagged relationships as well as other direct and indirect lagged effects were much the same. In comparison only the first cross-lagged relationship remained significant (12-stepyr1→ abstinenceyr5) for women (and two indirect paths were significant). Moreover, a noticeably lower effect size appeared for women in the relationship between index treatment and year-1 abstinence (β = 0.37 vs. 0.49 for men) and noticeably higher effect sizes appeared for contemporaneous relationships between 12-step attendance and abstinence at all follow-ups (0.45, 0.48. and 0.47 for women versus 0.39, 0.33 and 0.41 for men, respectfully) suggesting gender may moderate these relationships.
4. Discussion
4.1 Causal relationships
Our results follow with the prior studies that tested cross-lagged relationships with data more proximal to treatment initiation. Incorporating distal follow-up periods, our time-lagged results showed that greater 12-step attendance led to increases in 5-year abstinence and (to a lesser extent) in 7-year abstinence (gender differences are discussed below); causal associations in the reverse direction were not detected for those years. These cross-lagged relationships were obtained independent of baseline severity and treatment involvement effects. Importantly our analysis extends findings to a diverse population of treatment seekers, namely, women and men with alcohol and drug use disorders who were insured members of an integrated care organization that provides a range of outpatient treatment services like those in most other treatment programs in the United States (
Institute of Medicine (US), Institute of Medicine (US). Committee for the Study of Treatment, Rehabilitation Services for Alcoholism, Alcohol Abuse, National Institute on Alcohol Abuse, & Alcoholism (US).
Institute of Medicine Broadening the Base of Treatment for Alcohol Problems.
4.2 Effects of formal treatment duration on 12-step attendance and abstinence
Longer treatment duration led to a greater likelihood of past-30-day abstinence at year 1 and increased 12-step attendance in the last half of that year. In other words lengthier involvement in treatment had strong positive effects on short-term recovery efforts with this population. This suggests that longer treatment duration provides treatment-seekers with education, practice and counseling needed to start a recovery process. Moreover, the positive relationship between greater 12-step attendance following treatment initiation and abstinence as distally as 5 years later is consistent with other research indicating that initial heavy doses of post-treatment 12-step participation leads to better distal outcomes (
). Providers may find it beneficial to introduce amenable clients to the culture and practice of AA (or similar groups like Narcotics Anonymous) by way of 12-step facilitation interventions that bridge the gap between treatment and ongoing recovery efforts (
in: Recent Developments in Alcoholism: Research on Alcoholics Anonymous and Spirituality in Addiction Recovery. Vol. 18. Springer,
New York2008: 303-320
Twelve step facilitation therapy manual: A clinical research guide for therapists treating individuals with alcohol abuse and dependence [Project MATCH monograph series/DHHS publication, no. (ADM) 92-1893], Vol. 1.
Rockville, MD,
National Institute on Alcohol Abuse and Alcoholism1992
). While 12-step groups may not be the choice of all substance misusers (as suggested by our findings), they appear to be a valuable for option for many.
4.3 Causal linkages from index treatment to year-9 abstinence
The strongest time-lagged causal associations between treatment initiation and a greater likelihood of abstinence at 9 year were explained via three distinct and partially overlapping paths. Here we turn to a companion paper that used latent class trajectories analysis described above (
) in an attempt to more fully discuss and contextualize our path results. Unlike the path analysis, which was modeled to statistically test causative (time-lagged) relationships over time, the trajectories analysis clustered individuals based on their patterns of 12-step attendance over time and then compared the resultant classes on average rates of substance use at contemporaneous time points, hence, only associations (not causation) could be inferred.
The two indirect paths that led from treatment initiation via 12-step attendance and subsequent abstinence (years 5 and 7) exhibited patterns consistent with two attendance classes in the 2012 study; one dropped their attendance sharply and the other declined more gradually after year 1. Individuals in these classes reported lengthy treatment involvement and very high 12-step attendance in year 1 and they reported no or very low 12-step attendance at the 5-, 7- and 9-year follow-ups. Their reported abstinence rates (averaging over 60% at follow-ups) were second only to the class with the highest 12-step attendance at all follow-ups. Individuals in these two classes appeared to benefit greatly from their index treatment stay and their initial high dose of exposure to 12-step groups (like the two mentioned paths). Taken together, results under both analytical schemes reinforce the view that better long-term outcomes occur when preceded by high, early recovery involvement. The path analysis suggests causal effects.
In comparison, two other classes with very low and no attendance patterns reported the lowest average rates of abstinence at all follow-ups. These classes (nearly half total sample) were characterized by low treatment durations and lower baseline problem severity. We hypothesized that these treatment seekers may not have identified as well with those they met at 12-step meetings (or at treatment). Surprising, the average rate of abstinence for these classes was about 40% at follow-ups. Thus, some in these classes appear to have relied on other recovery resources more so than on treatment or 12-step programs and they look similar to those in the path connected by abstinence at all time-points. These individuals remain an interesting group to understand.
In direct contrast, a class with high 12-step attendance (abstinence rates averaging 75%) at all follow-ups (a quarter of the sample) that was characterized by more women than men and those with high baseline severity may well represent the path with successive 12-step attendance connections. Greater problem severity as represented here has been a reliable predictor of AA attendance (
Predictors of Alcoholics Anonymous Utilization among US Males and Females: A Longitudinal Analysis 35th Annual Alcohol Epidemiological Symposium of the Kettil Bruun Society for Social And Epidemiological Research on Alcohol, Copenhagen, Denmark: June 1–5, 2009. 2009
). Latent class trajectories and path analyses jointly suggest that women's abstinence status may be more contingent on concurrent 12-step attendance than is the case for men and even as distally as nearly a decade later. For both genders heavy first year recovery involvement was causally linked to long term abstinence.
4.4 Final remarks
We acknowledge some limitations. Foremost, these results will generalize best to other privately-insured populations. As is frequently the case, we rely on self-reported data for alcohol and drug abstinence. However, biological assays with these individuals tended to confirm self-reports (
). Moreover, we were limited to using past-30-day abstinence in our path models, but 30-day abstinence correlates strongly with longer periods of abstinence (
). We did not control for prior lifetime treatment episodes or prior 12-step attendance. Some evidence suggests that AA prior to treatment may provide a familiarity and a cumulative positive effect as related to post-treatment AA (
Alcoholics Anonymous involvement and positive alcohol-related outcomes: Cause, consequence, or just a correlate? A prospective 2-year study of 2,319 alcohol-dependent men.
Journal of Consulting and Clinical Psychology.2003; 71: 302-308
in: Recent developments in alcoholism: Research on Alcoholics Anonymous and spirituality in addiction recovery. Vol. 18. Springer,
New York2008: 357-372
). Although efforts were made to adjust attrition bias, low follow-up rates (69% at year 9) remain a limitation when interpreting our path model results and especially those for indirect effects. Last, our follow-up periods were clearly spaced too far apart to suggest direct causative effects: rather unexplained intervening factors likely influenced these associations.
In conclusion, our findings suggest that patients in integrated health care organizations benefit from 12-step attendance in much the same way other samples have benefited. As well, as behavioral health services move toward a recovery-oriented system of care (
Twelve-step affiliation and three-year substance use outcomes among adolescents: Social support and religious service attendance as potential mediators.
A longitudinal model of intake symptomatology, AA participation, and outcome: Retrospective study of the Project MATCH outpatient and aftercare samples.
in: Recent Developments in Alcoholism: Research on Alcoholics Anonymous and Spirituality in Addiction Recovery. Vol. 18. Springer,
New York2008: 303-320
Attendance at Narcotics Anonymous and Alcoholics Anonymous meetings, frequency of attendance and substance use outcomes after residential treatment for drug dependence: A 5 year follow-up study.
([URL:http://www.samefacts.com/2010/08/drug-policy/science-supports-the-12-steps-for-addiction/Accessed: 2010-11-01. (Archived by WebCite® at http://www.webcitation.org/5tv8pXory)])The reality-based community. 2010
Institute of Medicine (US), Institute of Medicine (US). Committee for the Study of Treatment, Rehabilitation Services for Alcoholism, Alcohol Abuse, National Institute on Alcohol Abuse, & Alcoholism (US).
Institute of Medicine Broadening the Base of Treatment for Alcohol Problems.
([URL:http://www.smh.com.au/opinion/society-and-culture/aa-cure-for-addicts-lacks-good-evidence-20100810-11y4b.html. Accessed: 2010-10-27. (Archived by WebCite® at http://www.webcitation.org/5tnZoPn98)]) The Sydney Morning Herald,
Sydney, NSW, Australia2010
([URL:http://www.washingtonpost.com/wp-dyn/content/article/2010/08/06/AR2010080602660.html. Accessed: 2010-10-27. (Archived by WebCite® at http://www.webcitation.org/5tnawLmcq)]) The Washington Post,
Washington, DC2010
Epidemiology of Alcoholics Anonymous participation.
in: Recent developments in alcoholism: Research on Alcoholics Anonymous and spiritual aspects in addiction recovery. Vol. 18. Springer,
New York2008: 261-282
Review of temporal effects and outcome predictors in substance abuse treatment studies with long-term follow-ups: Preliminary results and methodological issues.
Alcoholics Anonymous involvement and positive alcohol-related outcomes: Cause, consequence, or just a correlate? A prospective 2-year study of 2,319 alcohol-dependent men.
Journal of Consulting and Clinical Psychology.2003; 71: 302-308
Twelve step facilitation therapy manual: A clinical research guide for therapists treating individuals with alcohol abuse and dependence [Project MATCH monograph series/DHHS publication, no. (ADM) 92-1893], Vol. 1.
Rockville, MD,
National Institute on Alcohol Abuse and Alcoholism1992
in: Recent developments in alcoholism: Research on Alcoholics Anonymous and spirituality in addiction recovery. Vol. 18. Springer,
New York2008: 357-372
Predictors of Alcoholics Anonymous Utilization among US Males and Females: A Longitudinal Analysis 35th Annual Alcohol Epidemiological Symposium of the Kettil Bruun Society for Social And Epidemiological Research on Alcohol, Copenhagen, Denmark: June 1–5, 2009. 2009
☆This secondary analysis was supported by grants from the National Institute on Alcohol Abuse and Alcoholism (RO1AA10359) and the National Institute on Drug Abuse (R37DA10572 and R01DA08728).
00263-8&id=gr1.jpg){kind=link}