Comorbidity Of Substance Dependence And Depression

Psychology of Addictive Behaviors 2008, Vol. 22, No. 1, 47–57

In the public domain DOI: 10.1037/0893-164X.22.1.47

Comorbidity of Substance Dependence and Depression: Role of Life Stress and Self-Efficacy in Sustaining Abstinence Susan R. Tate

Johnny Wu

Veterans Affairs San Diego Healthcare System and University of California, San Diego

University of California, San Diego

John R. McQuaid

Kevin Cummins

Veterans Affairs San Diego Healthcare System and University of California, San Diego

University of California, San Diego

Chris Shriver and Marketa Krenek

Sandra A. Brown

Veterans Affairs San Diego Healthcare System

Veterans Affairs San Diego Healthcare System and University of California, San Diego

The authors examined life stress and self-efficacy as predictors of time to relapse for 113 adults with comorbid major depressive disorder and alcohol and/or substance dependence in a randomized clinical trial comparing 2 psychotherapy interventions (integrated cognitive– behavioral therapy and 12-step facilitation therapy). Life stress, self-efficacy, and substance use were assessed at treatment entry, 12 weeks (mid-treatment), and 24 weeks (end of treatment). Time to relapse was defined as the number of days from treatment initiation until first alcohol and/or drug use. Half of the sample relapsed within the study period of 24 weeks. There was no significant difference between treatment groups. Individuals experiencing life stressors were more likely to relapse early than those not experiencing life stressors. Lower self-efficacy also predicted earlier relapse. Chronic stress levels and self-efficacy were stable across time for most individuals. In contrast, acute stress events occurred at differing times, and survival analyses provided evidence of heightened relapse risk in the month following acute stressors. The interaction of self-efficacy and life stress was not significant. The results highlight the significance of life stress and self-efficacy as predictors of early relapse. Keywords: alcohol dependence, substance dependence, life stress, self-efficacy, survival analysis

Given the high prevalence of comorbid substance use and depressive disorders (Regier et al., 1990), it is important to under-

stand factors associated with treatment outcomes for individuals with these concomitant disorders. Individuals diagnosed with both depression and alcohol dependence have exhibited poorer drinking outcomes than alcohol-dependent individuals without depression (Greenfield et al., 1998). The presence of comorbid depression has also been shown to predict earlier relapse to alcohol and drug use among adolescents with alcohol or substance dependence (Cornelius et al., 2004). Higher levels of depressive symptoms have also predicted earlier treatment attrition, greater urges to use substances, and alcohol relapse (R. A. Brown et al., 1998). These findings highlight challenges associated with addiction treatment in the context of depressive disorders. Findings are mixed, however, with a number of studies not detecting different addiction outcomes for substance-dependent adults with and without depression (Carroll, Nich, & Rounsaville, 1995; Charney, Paraherakis, Negrete, & Gill, 1998; O’Sullivan et al., 1988; Sellman & Joyce, 1996; Tate, Brown, Unrod, & Ramo, 2004). A significant complication for dual-diagnosis treatment is the potential for alcohol or drug use to undermine psychotherapeutic and pharmacological intervention efforts for both disorders. Alcohol and drug use can cause or exacerbate depression symptoms, either through direct effects (e.g., alcohol, sedatives) or during withdrawal states (e.g., cocaine, amphetamines). Substance use can also compromise the effectiveness of pharmacotherapy inter-

Susan R. Tate and John R. McQuaid, Veterans Affairs San Diego Healthcare System, and Department of Psychiatry, University of California, San Diego; Johnny Wu, Department of Psychology, University of California, San Diego; Kevin Cummins, Department of Psychology and Department of Psychiatry, University of California, San Diego; Chris Shriver and Marketa Krenek, Veterans Affairs San Diego Healthcare System; Sandra A. Brown, Veterans Affairs San Diego Healthcare System, and Department of Psychology and Department of Psychiatry, University of California, San Diego. Johnny Wu is now at the Department of Psychology, University of Washington; Marketa Krenek is now at the Department of Psychology Syracuse University. This research was supported by a Veterans Affairs Medical Research Merit Review Grant awarded to Sandra A. Brown and a Veterans Affairs Medical Research Merit Review Entry Program Grant to Susan R. Tate. This study was completed as an honors thesis in the psychology honors program by Johnny Wu at University of California, San Diego, supervised by Susan R. Tate and Sandra A. Brown. Correspondence concerning this article should be addressed to Sandra A. Brown, Department of Psychology (0109), University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0109. E-mail: [email protected] 47

48

TATE ET AL.

ventions by causing serious side effects, potentiating effects of some psychotropic medications (thereby increasing risk of overdose), and decreasing adherence to medication regimens (Catz, Heckman, Kochman, & DiMarco, 2001; National Institute on Alcohol Abuse and Alcoholism, 2005). Finally, alcohol and substance use are associated with increases in suicidal ideation and suicide attempts, shifting intervention efforts to crisis management rather than other therapeutic goals (e.g., Claassen et al., 2007; Goldstein & Levitt, 2006; Shen et al., 2006). In addition to depression symptoms, numerous studies have demonstrated a relationship between life stress and worse posttreatment drinking outcomes (Billings & Moos, 1983; S. A. Brown, Vik, Patterson, Grant, & Schuckit, 1995; Canton et al., 1988; Vuchinich & Tucker, 1996). The predictive quality of life stressors also extends to posttreatment outcomes for cocaine (McMahon, 2001) and opiate users (Grey, Osborn, & Reznikoff, 1986). The negative impact of life stress may be particularly relevant for substance-dependent individuals with mood disorders, as life stress has also been associated with recurrence of depressive episodes, depression treatment outcomes, and attrition (e.g., Monroe, Kupfer, & Frank, 1992; Monroe, Roberts, Kupfer, & Frank, 1996). Indeed, some research supports the stress-generation hypothesis, which proposes that individuals with depression generate additional stressors as a result of their symptoms, behaviors, and social interactions (Hammen, 1991; Monroe, Slavich, Torres, & Gotlib, 2007), thus heightening risk of substance relapse. Evaluating the impact of life stress on outcomes is complex, however, given fluctuations in stress over time. In addition, the temporal nature of a stressor may affect risk of relapse. Stressors differ in their temporal characteristics, with some stressors persisting for extended periods, while other stressors occur on a specific date and are more short lived (Shiffman, 1989). Although much evidence has demonstrated a relationship between discrete stressful life events and substance use, chronic stressors have received much less attention. Our research has shown, however, that chronic stressors also increase the risk of resuming posttreatment substance use (S. A. Brown et al., 1990; Tate, McQuaid, & Brown, 2005). Self-efficacy is another personal characteristic that changes over time (Finney, Noyes, Coutts, & Moos, 1998; Rychtarik, Prue, Rapp, & King, 1992). Thus, the relationship between self-efficacy and addiction relapse may be most appropriately evaluated in models that consider such temporal changes. In the context of substance abuse, self-efficacy is defined as an individual’s belief in his or her ability to resist the urge to drink or use. Cognitive– behavioral models of addiction relapse postulate that those with high self-efficacy in their ability to abstain from alcohol and drugs are more likely to use coping responses and less likely to drink or use than those with low self-efficacy. Past research supports this relationship between self-efficacy and treatment outcomes for alcohol (e.g., Greenfield et al., 2000; Rychtarik et al., 1992; Sitharthan & Kavanagh, 1990), marijuana (e.g., Stephens, Wertz, & Roffman, 1993), and cocaine (e.g., Avants, Margolin, & Kosten, 1996). Self-efficacy may be particularly important for substancedependent individuals with comorbid depression, as depression may stem, in part, from conditions that lead individuals to believe that they are not able to successfully execute behaviors required to manage prospective situations (i.e., low self-efficacy; Bandura, 1982). Consistent with this conceptualization, research has linked

higher levels of depression symptoms to lower self-efficacy (Haukkala, Uutela, Vartiainen, McAlister, & Knekt, 2000; Kanfer & Zeiss, 1983). In a dual-diagnosis sample, individuals with greater psychiatric distress were also more tempted to drink (Velasquez, Carbonari, & DiClemente, 1999), which highlights the importance of self-efficacy for those with comorbidity. Thus far, we have discussed self-efficacy and life stress separately. The strength of an individual’s self-efficacy conviction is proposed to influence persistence in the face of obstacles and aversive life experiences (Bandura, 1992). Addiction relapse models suggest that individuals experiencing life stressors may remain abstinent in part because of their self-efficacy in being able to resist the urge to drink or use (Witkiewitz & Marlatt, 2004). This suggests the possibility that life stress and self-efficacy may jointly affect the relapse process. However, we found no studies exploring this potential interaction in comorbid samples of substance use– disordered individuals. Survival analysis is a technique that evaluates the time it takes for an event to happen and has been widely used for examining time to relapse (e.g., Brecht, von Mayrhauser, & Anglin, 2000; Cornelius et al., 2004; Greenfield et al., 2000; Jones & McMahon, 1994; Saunders, Baily, Phillips, & Allsop, 1993). All of these studies, however, used either unchanging predictors (e.g., gender, ethnicity) or potentially changing variables measured at a single time point (e.g., intake alcohol expectancies, intake self-efficacy). Many predictors, including life stress and self-efficacy, are dynamic and likely alter risk of relapse over time. Survival analysis is well suited for analyzing such time-varying predictors (Hosmer & Lemeshow, 1999; Willett & Singer, 1993; Willett, Singer, & Martin, 1998). For example, Hillegers et al. (2004) used survival analyses to examine the relationship between stressful life events as a time-varying predictor and the subsequent development of mood disorders in adolescents. Among other applications, Willett and Singer (1993) provided an example of survival techniques applied to cocaine relapse (Havassy, Hall, & Wasserman, 1991) using multiple assessments of mood as a time-varying predictor. The survivor function depicts the proportions of the sample who do not experience a given event—in this case, substance use—in each time period. We employ a Cox (1972) proportional-hazards model to determine whether time to relapse is influenced by life stress, self-efficacy, or their interaction. Our study is unique in that we have a comorbid sample with major depression and alcohol or substance use disorder, and we have modeled the change in predictors over time. We hypothesize that (a) severe life stress (both chronic stressors and discrete life events) will be associated with shorter survival times and (b) higher self-efficacy will be associated with longer survival times. In addition to our primary hypotheses, we also test whether the interaction between self-efficacy and life stress influences time to relapse beyond the unique contribution of these factors.

Method Participants Participants were patients at the Veterans Affairs San Diego Healthcare System, drawn from sequential referrals to the Substance Abuse Mental Illness Program, an abstinence-based outpatient dual-diagnosis clinic. Participants were included in the study

SELF-EFFICACY, STRESS, AND TIME TO RELAPSE

if they met current criteria for Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM–IV; American Psychiatric Association, 1994) major depressive disorder and alcohol, cannabis, or stimulant dependence with recent substance use (within 3 months prior to intake) and accepted the program goal of abstinence from alcohol and drug use. Participants were excluded if they (a) met criteria for DSM–IV bipolar disorder or any psychotic disorder, (b) met current criteria for DSM–IV opiate dependence through intravenous administration, (c) could not accurately recall events because of memory deficits, or (d) lived too far away to attend psychotherapy appointments twice a week. Participants agreed to (a) randomization to one of two psychotherapy groups; (b) faceto-face research assessments at intake, mid-treatment (12 weeks), and end of treatment (24 weeks); and (c) random toxicology screens. Additionally, all participants received pharmacotherapy via a standardized Veterans Affairs protocol for major depression and agreed not to participate in any other formal treatment for substance dependence or depression during the 24 weeks of treatment, with the exceptions of community 12-step meetings and residential treatment program required meetings. Ninety percent of consecutive referrals who met study criteria consented to participate in the study. Of those who refused consent, 1 person felt the assessments were overwhelming, 1 person expressed a preference to receive treatment focused on his comorbid anxiety disorder, and the remainder refused randomization to treatment condition. A total of 168 veterans gave informed consent to participate and were randomized to one of the two interventions. Thirty-two participants (19.0%) who did not complete their intake assessment were not included in the analyses: Six did not attend any therapy sessions (1 moved out of the area, and the others were not able to be contacted), and 26 participants attended at least one session but were not responsive to outreach efforts by either therapists or research staff. The remaining 136 participants were evaluated throughout treatment. One participant gave informed consent but refused further study participation at the time of the first research assessment, and 1 participant was deceased prior to completing treatment. Twenty-one additional participants (15.0%) were excluded from analysis because of missing mid-treatment data. The final sample included 113 individuals, with an average age of 48.9 years (SD ⫽ 7.4). The majority were male (94.7%) and Caucasian (72.6%) and had completed high school or the equivalent (96.5%). Most participants met lifetime DSM–IV criteria for alcohol dependence (90.3%), slightly more than half met criteria for stimulant dependence (54.8%), and 29.2% met criteria for marijuana dependence. Included and excluded cases did not significantly differ on demographic or dependence characteristics.

Design and Procedure Participants in this study were enrolled in a randomized clinical trial comparing integrated cognitive– behavioral therapy (ICBT) and 12-step facilitation therapy (TSF) for comorbid substance dependence and depression (S. A. Brown et al., 2006). Both interventions were delivered in a group format and consisted of two consecutive 12-week phases. Phase 1 consisted of 1-hr group sessions that met twice weekly for a total of 24 sessions from Weeks 1 through 12, and Phase 2 consisted of 1-hr group sessions that met once a week for a total of 12 sessions from Weeks 13 through 24. ICBT combined the interventions and structure of

49

cognitive– behavioral depression treatment (Mun˜oz, Ying, PerezStable, & Miranda, 1993) and the cognitive– behavioral coping skills training of Project MATCH (Kadden et al., 1994). ICBT focused on altering dysfunctional cognitions, practicing positive activities, and developing interpersonal communication skills. TSF consisted of the National Institute on Alcohol Abuse and Alcoholism Project MATCH TSF intervention (Nowinski, Baker, & Carroll, 1994), modified to be delivered in a group format rather than individual sessions. In addition to the study treatment groups, participants in both interventions also received monthly medication management appointments with the dual-diagnosis program psychiatrist, who used standardized Veterans Affairs protocol for major depressive disorder (e.g., selective serotonin reuptake inhibitors and atypical antidepressants). Within 1 week of obtaining informed consent for study participation, a trained research assistant completed a diagnostic assessment using the Composite International Diagnostic Interview (Robins et al., 1988) to confirm inclusion criteria. Current analyses are focused on assessments of self-efficacy, life stress, and substance use conducted at three interview time points: (a) intake, (b) mid-treatment (Week 12, end of Phase 1), and (c) end of treatment (Week 24, end of Phase 2).

Measures Self-efficacy. The 50-item self-report Drug-Taking Confidence Questionnaire (DTCQ; Annis & Martin, 1985) assesses coping self-efficacy for both alcohol and drug use and is a reliable and valid indicator of self-efficacy (alcohol sample, ␣ ⫽ .98; cocaine sample, ␣ ⫽ .98; Sklar, Annis, & Turner, 1997). Reliability coefficients were comparable in our sample (intake, ␣ ⫽ .98; mid-treatment, ␣ ⫽ .98). Participants rated their confidence in their ability to resist the urge to drink alcohol or use substances on a 0 –5 scale (0 ⫽ not at all confident to 5 ⫽ very confident) in 50 high-risk relapse situations. Depression symptoms. The Hamilton Depression Rating Scale (HDRS; Hamilton, 1960) is a widely used structured clinical interview that assesses depression symptoms experienced over the prior week. The HDRS consists of 21 items scored on a 0 – 4 scale, with demonstrated sensitivity and specificity in alcohol-dependent populations (Willenbring, 1986). Standards vary, but scores higher than 20 are considered indicative of clinically depressed patients (S. A. Brown et al., 1994). We report both mean scores and percentages above and below the cutoff at mid-treatment. Life stress. A trained interviewer administered the Psychiatric Epidemiology Research Interview—Modified (Hirschfield et al., 1977) at each assessment. This measure is a 133-item stressor checklist followed by a semistructured interview. For each stressor reported, the interviewer probed for a detailed description of the experience, including date of occurrence, context, duration, consequences, and proximity to substance use. The interviewer then presented this information to a panel of at least two raters who had achieved adequate reliability. The panel rated stressors according to the objective rating criteria of the Bedford College Life Events and Difficulties Schedule (G. W. Brown, Bifulco, Harris, & Bridge, 1986). Two coinvestigators on the study who had extensive prior experience with this rating system rated stressors and trained new raters.

50

TATE ET AL.

Following the training, new raters first observed and later participated in rating sessions with at least two previously trained raters, using the actual data from the study, until reliability was achieved (r ⬎.90). On the basis of the detailed presentation by the interviewer of each stressor, raters determined whether an occurrence qualified as a difficulty (ongoing chronic stressor lasting at least 4 weeks) or a stressful event (acute stressor with a discrete onset; G. W. Brown & Harris, 1978). All chronic difficulties and acute events were rated on the likelihood that the stressor would seriously threaten an individual’s personal or social well-being (Dohrenwend & Dohrenwend, 1974). Raters classified reported stressors (both chronic difficulties and acute events) as severe, nonsevere, or not posing adequate objective threat to warrant rating as a stressor on the basis of the standardized Bedford College Life Events and Difficulties Schedule manual (covering 516 pages of stressor examples and ratings). Because life stress can be the result of recent substance use (rather than preceding substance use), raters also coded whether each stressor was a result of recent substance use. Substance-induced stressors were excluded from analyses. In addition, we excluded health stressors, because prior research has conceptualized health problems as motivating abstinence rather than relapse, as is the case for other stressors (Monti et al., 1999; Smith, Hodgson, Bridgeman, & Shepherd, 2003), and we have demonstrated that health stressors are associated with reduced likelihood and severity of substance relapse (e.g., Tate et al., 2005). Substance use. The Timeline Follow-Back (TLFB; Sobell & Sobell, 1992) was used to measure all substance use during the 3 months prior to each assessment, including the date of first use following treatment initiation. We selected the date of first use for our outcome because of the importance of abstinence in dually diagnosed samples, noted previously, and because the goals and interventions of our dual-diagnosis program are abstinence based. The TLFB measures the type of substance the participant used and the number of days he or she used drugs as well as the quantity and frequency of alcohol use (Ehrman & Robbins, 1994; Fals-Stewart, O’Farrell, Freitas, McFarlin, & Rutigliano, 2000). The TLFB has demonstrated reliability and validity in alcohol- and substance-dependent (FalsStewart et al., 2000; Maisto, Sobell, & Sobell, 1979) and comorbid (Carey, 1997; Carey, Carey, Maisto, & Henson, 2004) treatment samples. Additionally, random toxicology screens were used to enhance reliability for participants’ self-reports of substance use. Sixtysix percent of participants had a urine toxicology screen at some time in Phase 1 or 2, and toxicology results matched TLFB daily selfreports 82.2% of the time (excluding one toxicology note of an inadequate sample). Two participants reported alcohol use that was not detected by the toxicology screen. Five participants denied alcohol or drug use at any time on the TLFB, but the toxicology screen indicated use (2 used marijuana, 3 used stimulants). An additional 12 participants reported use and had toxicology results indicating use, but the dates of use differed for self-report and toxicology results. If either source (self-report or toxicology) indicated use, use was coded for analyses. If both sources indicated use, the earlier of the two dates was used.

Statistical Analysis Time to relapse was analyzed with a Cox proportional-hazards model performed with treatment type, self-efficacy, chronic diffi-

culties, and acute events as predictor variables. As previously noted, survival analysis is widely used to analyze time-to-event data (e.g., Singer & Willett, 1993), and this methodology has been extended to handle time-varying covariates (Hosmer & Lemeshow, 1999). All analyses were conducted with STATA statistical software (Version 9.1; StataCorp., College Station, TX). The study period was predetermined to last until the end of treatment for each participant (range ⫽ 24 –27 weeks). Participants who did not relapse during the study period were fixed-right censored (Singer & Willett, 1993). A relapse was defined as the day an individual first used any amount of alcohol or drugs (Jones & McMahon, 1994). Survival time was measured as the number of days until initial substance use, beginning from each participant’s date of treatment initiation. An individual’s substance use status was coded for each day in our sample (abstinence vs. substance use). Although we did not anticipate treatment group differences in outcomes on the basis of our previous analyses (S. A. Brown et al., 2006), we included treatment groups (ICBT and TSF) in the model to control for possible differential treatment effects on time to relapse. We examined self-efficacy as a time-variant covariate because it was assumed to change over time as a result of treatment. We calculated the item mean self-efficacy score (average score across DTCQ items) using the intake assessment to predict survival time in Phase 1 and the mid-treatment scores to predict survival time in Phase 2. Two qualitatively distinct stress measures were included in our model. Chronic difficulties in each stressor domain (e.g., financial, work, relationship) were scored as none ⫽ 0, nonsevere ⫽ 1, and severe ⫽ 2. For each phase, difficulty scores in each domain were summed to provide an index of overall chronic stress. The statistical distribution of the summed indexes was highly skewed, with the majority of participants scoring 1–2. The limited research examining difficulties has commonly dichotomized the presence or absence of a difficulty in samples with lower levels of chronic difficulties (e.g., Monroe et al., 2007). Given the high levels of chronic difficulties in our sample, we trichotomized each participant’s index such that 0 indicated having no difficulties (summed index ⫽ 0; 13.3% of the sample at intake), 1 indicated low chronic difficulty levels (summed indexes ⫽ 1–2; 58.4% at intake), and 2 indicated high chronic difficulty levels (summed indexes 3–7; 28.3% at intake). Like self-efficacy, chronic difficulties varied by phase of treatment (i.e., having a score for both Phase 1 and Phase 2). The second stress measure included in our model represented acute events. The acute event measure was dichotomously coded for absence or presence of an acute event. As our prior research has documented that substance relapse is related to acute events rated as severe (S. A. Brown, Vik, Patterson, Grant, & Schuckit, 1995), we excluded nonsevere events from our analysis. We assumed that if a relapse was related to an acute event, the relapse would likely occur within 30 days following the event (Tate, 2000). Thus, a participant was coded as having exposure to an acute event for a 30-day window following onset of each acute event. Acute events, like substance use, were not constrained to changing at the midtreatment assessment but could change on any day in the time period. In summary, survival analysis involves computing the proportion of a group that experiences the event of interest in a given period—in our study, on a daily basis. The survivor

SELF-EFFICACY, STRESS, AND TIME TO RELAPSE

function is a plot of these daily proportions accumulated across time. One can compute daily risks to examine whether risk (or hazard) differs systematically for different groups within a sample. As described by Willett and Singer (1993), survival analyses resemble logistic regression models with daily abstinence or relapse as the outcome variable and model variables as predictors accumulated over time. As separate proportions are calculated for each day, both static predictors (ICBT vs. TSF treatment group) and time-varying predictors (self-efficacy, chronic difficulties, acute events) can be included, coded on a day-to-day basis. Finally, the interaction of life stress and self-efficacy was evaluated in an extended model. Two interaction terms were added to the initial Cox proportional-hazards model: Self-Efficacy ⫻ Chronic Difficulties and SelfEfficacy ⫻ Acute Events. Thus, we evaluated the ongoing risk of relapse given an individual’s treatment (ICBT vs. TSF), current self-efficacy, acute events, and chronic difficulties and the interaction of self-efficacy and life stress.

Results Substance Use Outcomes Fifty-six (50%) individuals relapsed within the study period, 38 (34%) reported no alcohol or drug use, and 19 (17%) reported no alcohol and drug use in Phase 1 and were censored because of missing Phase 2 outcome or predictor variables (i.e., 17 were missing mid-treatment DTCQs, and 2 were missing Phase 2 TLFBs). There were no significant differences between the 19 censored and 94 remaining cases on any demographic or model variables. Most individuals who relapsed did so in the first 60 days

51

(39% in the 1st month, and 30% in the 2nd month). The remaining 31% of the relapses were spread across the remainder of the study period. Of the 56 relapses, 31 individuals (55%) drank alcohol, 8 (14%) used stimulants (i.e., cocaine or methamphetamine), 8 (14%) used marijuana, and 9 (17%) reported multiple substances used. All participants who relapsed used a substance for which they had met diagnostic criteria.

Depression Symptoms and Substance Use Sample characteristics by relapse status are depicted in Table 1. Although it is not a focus of this study and was not included in the model, depression was examined as a predictor of relapse in each phase via logistic regressions. Depression symptoms were not related to likelihood of relapse in either phase; intake HDRS predicting relapse in Phase 1, ␹2(1, N ⫽ 113) ⫽ 0.14, p ⫽ .71, odds ratio ⫽ 1.01; mid-treatment HDRS predicting relapse in Phase 2, ␹2(1, N ⫽ 49) ⫽ 0.21, p ⫽ .66, odds ratio ⫽ 0.99. Depression scores for 45% of the participants were below the HDRS clinically depressed cutoff of 20 at the mid-treatment assessment. Remission of depression symptoms at midtreatment was not related to relapse in Phase 2, ␹2(1, N ⫽ 49) ⫽ 0.002, p ⫽ .97 (22.7% of participants below the cutoff relapsed, and 22.2% of participants above the cutoff relapsed).

Model Predicting Time to Relapse The Cox proportional-hazards model was used to predict time to relapse via treatment type, self-efficacy, chronic difficulties, and acute events. Covariates were first examined for both proportion-

Table 1 Sample Characteristics by Relapse Status in Phases 1 and 2

Characteristic Treatment (% TSF / % ICBT) Mean (SD) depression Intake Mid-treatment Depression (% in remission)b Mid-treatment Mean (SD) self-efficacy Intake Mid-treatment Self efficacy (% low/middle/high) Intake Mid-treatment Chronic difficulties (% none/low/high) Phase 1 Phase 2 Acute events (%) Phase 1c Phase 2

Survivors (abstainers) (n ⫽ 38)

Phase 1 relapse (n ⫽ 45)

Phase 2 relapse (n ⫽ 11)

Censoreda (missing data) (n ⫽ 19)

45/55

44/56

46/55

63/37

28.6 (12.0) 23.2 (13.5)

29.1 (12.4) 27.3 (13.3)

25.3 (10.9) 21.2 (11.8)

29.3 (12.4) 27.5 (10.6)

45

26

46

20

3.5 (1.0) 4.0 (1.0)

3.0 (1.2) 3.2 (1.1)

3.6 (1.1) 4.0 (0.8)

3.5 (1.2) 3.8 (1.5)

0/42/58 0/24/76

13/40/47 3/47/50

9/18/73 0/18/82

5/37/58 0/50/50

16/60/24 21/55/24

13/47/40 16/51/33

0/73/27 0/80/20

16/74/10 6/76/18

24 5

21 18

9 36

11 26

Note. TSF ⫽ twelve-step facilitation therapy; ICBT ⫽ integrated cognitive-behavioral therapy. a The censored group was composed of survivors (abstainers) in Phase 1 who were excluded because of missing data in Phase 2. Mid-treatment self-efficacy scores were available for only 2 participants, and mid-treatment depression scores were available for 10 participants in this group. bHamilton Depression Rating Scale scores less than 20 were coded as remitted. cNineteen percent of Phase 1 survivors experienced an acute event in the 12 weeks of Phase 1 (across survivor and censored groups). In contrast, 21% of the Phase 1 relapse group experienced an acute life event prior to relapse (M ⫽ 32 days, SD ⫽ 24 days, Mdn ⫽ 30 days).

TATE ET AL.

52

ality of hazards and multicollinearity. None of the covariates significantly interacted with time, and there was no evidence of multicollinearity. Time to relapse was predicted by the model, ␹2(4, N ⫽ 113) ⫽ 20.44, p ⬍ .001. Type of treatment received was not significantly associated with risk of substance relapse during treatment in this comorbid sample (hazard ratio ⫽ 1.13, 95% confidence interval [CI] ⫽ 0.67–1.92, p ⫽ .65; see Figure 1a). Median survival times were 148 days for TSF (n ⫽ 54; 48% of the sample) and 122 days for ICBT (n ⫽ 59; 52% of the sample). Table 1 shows sample characteristics by relapse status, and Table 2 provides the median survival days (time to relapse) for participants by treatment, selfefficacy level, and life stressors (chronic difficulties and acute events).

Self-Efficacy and Time to Relapse We assumed self-efficacy would change over time as a result of treatment, and DTCQ scores increased significantly from the intake assessment to the Phase 1 assessment at mid-treatment, t(84) ⫽ 2.98, p ⫽ .004. Self-efficacy was a significant predictor in the model ( p ⫽ .003). Each unit increase in the DTCQ decreased the risk of relapse by a hazard ratio of 0.71 (95% CI ⫽ 0.56 – 0.89).

Table 1 lists DTCQ scores by relapse status for each phase. In Phase 2, little difference was seen between self-efficacy scores by relapse status, as those with low self-efficacy relapsed early (see Figure 1b). As continuous self-efficacy measures (either mean or total scores) are most common in the research literature (e.g., Blume, Schmaling, & Marlatt, 2001; Maisto, Clifford, Longabaugh, & Beattie, 2002), analyses were conducted with mean DTCQ scores. For presentation in the figure, self-efficacy was split into three levels to simplify display and interpretation (Singer & Willett, 1993). We produced survival functions by trichotomizing DTCQ scores: (a) low indicates individuals who had scores of 0.00 –1.66, (b) medium indicates individuals who had scores of 1.67–3.33, and (c) high indicates individuals who had scores of 3.34 –5.00. As shown in Figure 1b, individuals with higher levels of self-efficacy had longer survival times, and lower self-efficacy was predictive of relapse early in treatment. Table 2 lists the median survival days by these categories. Using the trichotomized categories for the 49 participants who survived to the second phase, we found that 33 participants (67.3%) did not change self-efficacy levels (26 started and remained high, 7 started and remained in the middle category, and no one started and remained in the low category) and 16 individuals (32.7%) did change self-

Figure 1. Estimated survivorship functions following treatment initiation as related to (a) treatment type, (b) self-efficacy, (c) chronic difficulties, and (d) acute events. TSF ⫽ 12-step facilitation therapy; ICBT ⫽ integrated cognitive– behavioral therapy.

SELF-EFFICACY, STRESS, AND TIME TO RELAPSE

Table 2 Survival Characteristics by Treatment, Self-Efficacy, Acute Events, and Chronic Difficulties Variable Treatment TSFb ICBTc Self-efficacyd Low Medium High Chronic difficulties None Moderate Severe Acute events None Any

No. participantsa

Median survival days

54 59

148 122

8 47 74

18 85 170

17 69 33

— 170 62

110 20

— 41

Note. Dashes indicate that fewer than half of the participants relapsed; hence, median survival time was beyond the study period. TSF ⫽ twelvestep facilitation therapy; ICBT ⫽ integrated cognitive-behavioral therapy. a Number of participants who were in each condition at least once during b c the study period. Twelve-step facilitation therapy. Integrated d cognitive– behavioral therapy. Self-efficacy scores were categorized into low (0.00 –1.66), medium (1.67–3.33), and high (3.34 –5.00) groups.

efficacy levels (4 decreased from the high to the middle level, 1 person increased from low to high, and 11 increased from middle to high). Twenty-four percent of those with unchanged selfefficacy relapsed in Phase 2, compared to 25% of participants with decreased self-efficacy and 17% of participants with increased self-efficacy. Of note, only 2 participants with low self-efficacy did not relapse in Phase 1: One person relapsed in Phase 2 despite a self-reported increase from low to high self-efficacy at midtreatment, and 1 person was censored because of missing Phase 2 data.

Stressors and Time to Relapse A range of acute events and chronic difficulties was experienced during the study period. Eighty percent of the total sample had a chronic financial difficulty during the study, 20% had a chronic legal difficulty, 20% had a chronic relationship difficulty, 19% had a chronic housing difficulty, and a single individual had a chronic work difficulty. Twelve percent of the sample had no chronic difficulty during the entire study period, 46% had one chronic difficulty, and 42% had multiple chronic difficulties. The domains of acute events experienced were as follows: Six percent of the total sample had a relationship event, 6% had a legal event, 4% had a housing event, 3% had a death event, 3% had a work event, and a single individual had a financial event. Table 1 lists the percentages of participants with chronic difficulties and acute events by relapse status. For the survivor and censored groups, the acute events percentage reflects participants who experienced an acute event at any time in Phase 1 (initial 12 weeks) or Phase 2 (subsequent 12 weeks). For the relapse groups, the percentage reflects participants who experienced an acute event prior to the day of relapse (Phase 1, M ⫽ 32 days, SD ⫽ 24 days; Phase 2, M ⫽ 122 days, SD ⫽ 31 days).

53

Both types of stress were significant predictors in the model. For chronic difficulties ( p ⫽ .02), each unit increase in the trichotomized stress score increased the risk of relapse by a hazard ratio of 1.66 (95% CI ⫽ 1.08 –2.55). Increasing levels of chronic difficulties shortened survival time (see Figure 1c). Among the 49 participants who did not relapse in Phase 1, the majority did not change chronic difficulty levels in Phase 2 (n ⫽ 43). Six individuals had trichotomized difficulty scores that changed levels at mid-treatment (1 increased and 4 decreased); none of these 6 participants relapsed in Phase 2. In the group with no chronic difficulties, fewer than half of the cases relapsed; hence, median survival time was beyond our study period (see Table 2). Acute events were also significant in the model ( p ⫽ .009), such that having an acute event increased the risk of relapse by a hazard ratio of 2.91 (95% CI ⫽ 1.30 – 6.51). As depicted in Figure 1d, individuals who were under the influence of an acute event (bottom function), compared to those who were not (top function), had shorter survival times. Because events are acute experiences occurring on a discrete date and we assumed increased risk could persist for 30 days (Tate, 2000), individuals fluctuated between functions depending on whether they had experienced an acute event within the prior 30 days. Thus, a participant’s risk of relapse was depicted by the “absent” curve until the day of an acute event. As of that date, risk of relapse was depicted by the “present” curve for the subsequent 30 days, at which time the risk returned to the decreased risk levels depicted in the absent curve. Five individuals reported multiple concurring acute events. The median survival time for individuals who did not experience an acute event was beyond our study period (see Table 2). Finally, the interactions of chronic difficulties with self-efficacy and acute events with self-efficacy were added to the initial model. The inclusion of the interaction terms did not significantly improve the model, ⌬␹2(2, N ⫽ 113) ⫽ 2.60, p ⫽ .27. Interpretation of this finding is guarded given the limited range of self-efficacy observed in our sample over the time frame, as the majority of participants with low self-efficacy relapsed early.

Discussion The present study demonstrates significant relationships between self-efficacy, life stress, and time to relapse during the initial 6 months in treatment for adults with comorbid major depressive disorder and substance dependence. Consistent with prior alcohol and drug research (Greenfield et al., 2000; Rychtarik et al., 1992; Stephens et al., 1993), we found that individuals with higher levels of self-efficacy maintained abstinence from alcohol and drugs longer than those with lower levels of self-efficacy. Indeed, a prominent feature of the survival function (Figure 1b) was the sharp decline in abstinence seen in those with low self-efficacy in the early weeks of treatment. The difference in survival between those with middle and high levels of self-efficacy was much less pronounced. Our findings vividly portray the heightened risk of relapse for those with low self-efficacy early in treatment. Speculation regarding the impact of changes in self-efficacy during treatment is limited by the fact that almost everyone with low self-efficacy in our sample relapsed very early in treatment. Fortunately, a minority of patients fell into the low self-efficacy range. Future development and testing of intensive interventions aimed at

54

TATE ET AL.

this small portion of high-risk individuals early in treatment may improve outcomes. Both chronic and acute stress elevated relapse risk among the dually diagnosed sample of this study. Those with higher levels of chronic difficulties relapsed sooner and were at greater risk of relapsing. High levels of chronic difficulties more than doubled the risk of relapse compared to individuals with no chronic difficulty. As chronic stress became increasingly severe and extended across multiple domains (e.g., financial, legal, and housing), the risk of relapse increased. Assessing such protracted stress identifies individuals at high risk for early return to drinking or using. Chronic life difficulties may contribute to increased likelihood of resuming substance use by posing more adaptational demands on the individual, through protracted exposure to negative affective states (e.g., depression, anxiety, and worry), or through depletion of coping resources (e.g., social support and financial resources). Substance abusers coping with chronic life stress may also be susceptible to developing a sense of learned helplessness when managing persistent problems with newly acquired coping skills that are perceived as more effortful and less immediately effective than substance use. Many addiction treatment programs incorporate stress management interventions focused on enhancing coping skills in high-risk situations. Our findings suggest that the inclusion of stress management interventions may be particularly useful for comorbid populations, given that the vast majority of our participants reported experiencing these protracted stressors. Additionally, these chronic stressors were stable over time for most of our sample. In contrast to discrete high-risk situations, which are often the focus of addiction stress management interventions, the chronicity of these difficulties highlights the need for coping skills training specific to managing prolonged stress. In addition to stress management skills training, interventions targeting specific types of stressors may be beneficial. For example, the most common type of difficulty in our study was in the financial domain, and vocational rehabilitation programs may be helpful for substance abusers experiencing problems in this domain. Providing other resources (e.g., legal or relationship counseling) as an adjunct to treatment may ameliorate or postpone the likelihood of turning to substance use for individuals coping with these types of chronic difficulties. Acute stress events nearly tripled the risk of relapse, significantly reducing survival times. This finding supports prior research demonstrating that these types of stressors are related to poorer substance use outcomes (S. A. Brown et al., 1995; Canton et al., 1988; McMahon, 2001; Vuchinich & Tucker, 1996). Despite recent studies using survival analysis to examine the relationship between stressful life events and depression (Hillegers et al., 2004; Kendler, Kuhn, & Prescott, 2004), our study is the first to examine the time-limited relationship between the occurrence of an acute stress event and relapse within 30 days following the event. Fortunately, acute events designated as severe on the Bedford College Life Events and Difficulties Schedule (Bifulco et al. 1989; G. W. Brown et al., 1986) life stress rating criteria tend to be infrequent occurrences (e.g., death of a loved one, divorce, eviction, and incarceration). In contrast to the protracted nature of negative affect associated with chronic difficulties, the intensity of emotional response to acute events may overwhelm substance abusers, which suggests distinct coping and intervention needs for comorbid populations during addiction treatment. The continuing decline

in the event curve over time suggests that acute events continued to pose risk throughout the treatment phases we examined. Temporary increases in the frequency of treatment provider contact (e.g., booster sessions) or stress-specific adjunct interventions, such as grief counseling, may be of benefit. Our findings do not support an interaction between self-efficacy and life stressors in predicting time to relapse. It is possible that we lacked adequate power to detect an interaction of this magnitude, or the complexities associated with these variables in our sample might have limited our ability to detect such moderating effects. For example, an individual’s self-efficacy estimates may reflect self-confidence in the context of chronic difficulties, which would reduce the likelihood of detecting an interaction. Additionally, since individuals with low self-efficacy relapsed early in our study, little variation in self-efficacy remained across later time periods, limiting statistical power for detecting interaction effects over the study period. Future studies with these variables are needed to ascertain whether the distributional characteristics of our predictor variables over time are replicated across settings. The likelihood of early relapse was similar across the two types of treatment in our study (ICBT vs. TSF). We had not anticipated differences on the basis of our prior analyses of substance use outcomes during treatment (percentage of days abstinent, average drinks per drinking day) in this sample (S. A. Brown et al., 2006). The addiction portions of our interventions were based on manuals developed for Project MATCH, and few differences were detected between the TSF and cognitive– behavioral coping skills conditions in that study (Project MATCH Research Group, 1998). Additionally, neither severity of depression symptoms nor depression remission predicted relapse within our sample of depressed substance-dependent adults. In some prior studies comparing depressed and nondepressed substance-dependent patients, depression predicted worse addiction outcomes (e.g., R. A. Brown et al., 1998; Cornelius et al., 2004; Greenfield et al., 1998). However, our findings suggest that variance in depression levels above clinical thresholds is not related to substance relapse. Of note, participants in our remitted group at mid-treatment were not asymptomatic but continued to experience symptoms. Depression research has suggested that residual depressive symptoms constitute a “subthreshold continuation of an active major depressive episode” (Judd, Paulus, Zeller, 1999, p. 764; Judd et al., 2000), with heightened risk for depression relapse. Thus, remitted participants in our sample are not comparable to nondepressed participants in prior studies. Our results should be interpreted in the context of potential sample and methodological limitations. These findings were based on a sample of predominantly White men and may not generalize to other groups. Our sample included substance-dependent individuals with comorbid depression, and future research is needed to determine whether the relationships between hypothesized predictors and time to relapse extend to other common types of comorbidity (e.g., anxiety disorders, schizophrenia, and bipolar disorder). Although our trichotomization of chronic difficulties appeared to capture qualitative distinctions among stress levels, more research is needed to validate this classification. In addition, larger studies should examine whether various domains of stress (e.g., relationship, financial, housing) differentially affect time to relapse. Our prior research suggested that relapse was most likely to occur within 30 days following a acute event, but an individual

SELF-EFFICACY, STRESS, AND TIME TO RELAPSE

may not be free of negative impact of the event after this period. More research is needed to examine the length of impact for acute stress events on subsequent substance use. As mentioned previously, we selected initial substance use as our outcome variable on the basis of the abstinence goals of our dual-diagnosis program and the potential for substance use to negatively impact dual-diagnosis treatment. However, addiction models posit different predictors for initial substance use episodes versus protracted use (e.g., Marlatt & Gordon, 1980), and our prior research has also detected differences in predictors for initiation versus continuation of posttreatment substance use (Tate, 2000; Tate et al., 2005). Thus, more studies are needed to examine the relationship of life stress and selfefficacy to other important substance outcomes. Finally, it was beyond the scope of this study to incorporate depression symptoms in our survival models. Future models including both depression and addiction outcomes in dual-diagnosis studies are needed. In summary, the present study highlights the importance of self-efficacy, chronic difficulties, and acute events on time to relapse for substance-dependent adults with depression. We have demonstrated that individuals with comorbid substance dependence and depression are at a heightened risk for relapse and earlier resumption of substance use when experiencing life stressors, both chronic difficulties and acute events. Consistent with prior research and theory, self-efficacy was also related to risk of resuming substance for comorbid adults. As previously noted, extending abstinence may be particularly important in treatment of individuals with substance use and other comorbid psychiatric disorders. Alcohol and substance use can undermine pharmacotherapy interventions through decreased medication adherence, missed appointments, and compromised evaluation of response to medication. Further, substance relapse early in the course of treatment will likely impair one’s ability to learn and effectively utilize new skills needed to address either disorder. Our findings clarify particular areas of vulnerability, aiding clinicians in identifying individuals at greater risk. Future research is needed to evaluate whether additional provider contact or targeted intervention efforts would prove beneficial in ameliorating the heightened risks posed by life stressors.

References American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Annis, H. M., & Martin, G. (1985). The Drug-Taking Confidence Questionnaire. Toronto, Ontario, Canada: Addiction Research Foundation of Ontario. Avants, S. K., Margolin, A., & Kosten, T. R. (1996). Influence of treatment readiness on outcomes of two pharmacotherapy trials for cocaine abuse among methadone-maintained patients. Psychology of Addictive Behaviors, 10, 147–156. Bandura, A. (1982). Self-efficacy mechanisms in human agency. American Psychologist, 37, 122–147. Bandura, A. (1992). Self-efficacy mechanism in psychobiologic functioning. In R. Schwarzer (Ed.), Self-efficacy: Thought control of action (pp. 355–394). Washington, DC: Hemisphere. Bifulco, A., Brown, G. W., Edwards, A., Harris, T., Neilson, E., Richards, C. & Robinson, R. (1989). Bedford College Life Events and Difficulties Schedule. London: University of London. Billings, A. G., & Moos, R. H. (1983). Psychosocial processes of recovery among alcoholics and their families: Implications for clinicians and program evaluators. Addictive Behaviors, 8, 205–218.

55

Blume, A. W., Schmaling, K. B., & Marlatt, G. A. (2001). Motivating drinking behavior change: Depressive symptoms may not be noxious. Addictive Behaviors, 26, 267–272. Brecht, M. L., von Mayrhauser, C., & Anglin, M. D. (2000). Predictors of relapse after treatment for methamphetamine use. Journal of Psychoactive Drugs, 32, 211–220. Brown, G. W., Bifulco, A., Harris, T. O., & Bridge, L. (1986). Life stress, chronic subclinical symptoms and vulnerability to clinical depression. Journal of Affective Disorders, 11, 1–19. Brown, G. W., & Harris, T. O. (1978). Social origins of depression: A study of psychiatric disorder in women. New York: Free Press. Brown, R. A., Monti, P. M., Myers, M. G., Martin, R. A., Rivinus, T., Dubreuil, M. E., et al. (1998). Depression among cocaine abusers in treatment: Relation to cocaine and alcohol use and treatment outcome. American Journal of Psychiatry, 155, 220 –225. Brown, S. A., Glasner, S. V., Tate, S. R., McQuaid, J. R., Chalekian, J., & Granholm, E. (2006). Integrated cognitive behavioral therapy versus twelve step facilitation therapy for substance dependent adults with depressive disorders. Journal of Psychoactive Drugs, 38, 449 – 460. Brown, S. A., Inaba, R. K., Gillin, J. C., Schuckit, M. A., Stewart, M. A., & Irwin, M. R. (1994). Alcoholism and affective disorder: Clinical course of depressive symptoms. American Journal of Psychiatry, 152, 45–52. Brown, S. A., Vik, P. W., McQuaid, J. R., Patterson, T. L., Irwin, M. R., & Grant, I. (1990). Severity of psychosocial stress and outcome of alcoholism treatment. Journal of Abnormal Psychology, 99, 344 –348. Brown, S. A., Vik, P. W., Patterson, T. L., Grant, I., & Schuckit, M. A. (1995). Stress, vulnerability and adult alcohol relapse. Journal of Studies on Alcohol, 56, 538 –545. Canton, G., Giannini, L., Magni, G., Bertinaria, A., Cibin, M., & Gallimberti, L. (1988). Locus of control, life events, and treatment outcome in alcohol dependent patients. Acta Psychiatrica Scandinavica, 78, 18 –23. Carey, K. (1997). Reliability and validity of the Time-Line Follow-Back interview among psychiatric outpatients: A preliminary report. Psychology of Addictive Behaviors, 11, 26 –33. Carey, K. B., Carey, M. P., Maisto, S. A., & Henson, J. M. (2004). Temporal stability of the Timeline Followback Interview for alcohol and drug use with psychiatric outpatients. Journal of Studies on Alcohol, 65, 774 –781. Carroll, K. M., Nich, C., & Rounsaville, B. J. (1995). Differential symptom reduction in depressed cocaine abusers treated with psychotherapy and pharmacotherapy. Journal of Nervous and Mental Disease, 183, 251– 259. Catz, S. L., Heckman, T. G., Kochman, A., & DiMarco, M. (2001). Rates and correlates of HIV treatment adherence among late middle-aged and older adults living with HIV disease. Psychology, Health & Medicine, 6(1), 47–58. Charney, D. A., Paraherakis, A. M., Negrete, J. C., & Gill, K. J. (1998). The impact of depression on the outcome of addictions treatment. Journal of Substance Abuse Treatment, 15(2), 123–130. Claassen, C. A., Trivedi, M. H., Rush, A. J., Husain, M. M., Zisook, S., Young, E., et al. (2007). Clinical differences among depressed patients with and without a history of suicide attempts: Findings from the STAR*D trial. Journal of Affective Disorders, 97, 77– 84. Cornelius, J. R., Maisto, S. A., Martin, C. S., Bukstein, O. G., Salloum, I. M., Daley, D. C., et al. (2004). Major depression associated with earlier alcohol relapse in treated teens with AUD. Addictive Behaviors, 29, 1035–1038. Cox, D. R. (1972). Regression models and life tables. Journal of the Royal Statistical Society, Series B, 34, 187–202. Dohrenwend, B. S., & Dohrenwend, B. P. (1974). Stress life events: Their nature and effects. New York: Wiley. Ehrman, R. N., & Robbins, S. J. (1994). Reliability and validity of 6-month

56

TATE ET AL.

timeline reports of cocaine and heroin use in a methadone population. Journal of Consulting and Clinical Psychology, 62, 843– 850. Fals-Stewart, W., O’Farrrell, T. J., Freitas, T. T., McFarlin, S. K., & Rutigliano, P. (2000). The Timeline Follow-Back reports of psychoactive substance use by drug-abusing patients: Psychometric properties. Journal of Consulting and Clinical Psychology, 68, 134 –144. Finney, J. W., Noyes, C. A., Coutts, A. I., & Moos, R. H. (1998). Evaluating substance abuse treatment process models: I. Changes on proximal outcome variables during 12-step and cognitive-behavioral treatment. Journal of Studies on Alcohol 59, 371–380. Goldstein, B. I., & Levitt, A. J. (2006). Is current alcohol consumption associated with increased lifetime prevalence of major depression and suicidality? Results from a pilot community survey. Comprehensive Psychiatry, 47, 330 –333. Greenfield, S., Hufford, M., Vagge, L., Muenz, L., Costello, M., & Weiss, R. (2000). The relationship of self-efficacy expectancies to relapse among alcohol dependent men and women: A prospective study. Journal of Studies on Alcohol, 61, 345–351. Greenfield, S. F., Weiss, R. D., Muenz, L. R., Vagge, L. M., Kelly, J. F., Bello, L. R., et al. (1998). The effect of depression on return to drinking: A prospective study. Archives of General Psychiatry, 55, 259 –265. Grey, C., Osborn, E., & Reznikoff, M. (1986). Psychosocial factors in outcome in two opiate addiction treatments. Journal of Clinical Psychology, 42, 185–189. Hamilton, M. (1960). A rating scale for depression. Journal of Neurological and Neurosurgical Psychiatry, 23, 56 – 62. Hammen, C. (1991). Generation of stress in the course of unipolar depression. Journal of Abnormal Psychology, 100, 555–561. Haukkala, A., Uutela, A., Vartiainen, E., McAlister, A., & Knekt, P. (2000). Depression and smoking cessation: The role of motivation and self-efficacy. Addictive Behaviors, 25, 311–316. Havassy, B. E., Hall, S. M., & Wasserman, D. A. (1991). Social support and relapse: Commonalities among alcoholics, opiate users, and cigarette smokers. Addictive Behaviors, 16, 235–246. Hillegers, M. H. J., Burger, H., Wals, M., Reichart, C. G., Verhulst, F. C., Nolen, W. A., et al. (2004). Impact of stressful life events, familial loading and their interaction on the onset of mood disorders. British Journal of Psychiatry, 185, 97–101. Hirschfield, R. M. A., Lerman, G. L., Schless, A. P., Endicott, J., Lichtenstaeder, S., & Clayton, P. J. (1977). Modified Life Events sections of the Psychiatric Epidemiology Research Interview (PERI-M). Bethesda, MD: National Institute of Mental Health. Hosmer, D. W., & Lemeshow, S. (1999). Applied survival analysis: Regression modeling of time to event data. New York: Wiley. Jones, B. T., & McMahon, J. (1994). Negative alcohol expectancy predicts post-treatment abstinence survivorship: The whether, when and why of relapse to first drink. Addiction, 89, 1653–1665. Judd, L. L., Paulus, M. J., Schettler, P. J., Akiskal, H. S., Endicott, J., Leon, A. C., et al. (2000). Does incomplete recovery from first lifetime major depressive episode herald a chronic course of illness? American Journal of Psychiatry, 157, 1501–1504. Judd, L. L., Paulus, M. P., & Zeller, P. (1999). The role of residual subthreshold depressive symptoms in early episode relapse in unipolar major depressive disorder. Archives of General Psychiatry, 56, 764 – 765. Kadden, K., Carroll, K., Donovan, D., Cooney, N., Monti, P., Abrams, D., et al. (1994). Cognitive-behavioral coping skills therapy manual. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism. Kanfer, R., & Zeiss, A. M. (1983). Depression, interpersonal setting, and judgments of self-efficacy. Journal of Abnormal Psychology, 92, 319 – 329. Kendler, K. S., Kuhn, J. W., & Prescott, C. A. (2004). Childhood sexual abuse, stressful life events and risk for major depression in women. Psychological Medicine, 24, 1475–1482.

Maisto, S. A., Clifford, P. R., Longabaugh, R., & Beattie, M. (2002). The relationship between abstinence for one year following pretreatment assessment and alcohol use and other functioning at two years in individuals presenting for alcohol treatment. Journal of Studies on Alcohol, 63, 397– 403. Maisto, S. A., Sobell, L. C., & Sobell, M. B. (1979). Comparison of alcoholics’ self-reports of drinking behavior with reports of collateral informants. Journal of Consulting and Clinical Psychology, 47, 106 – 122. Marlatt, G. A., & Gordon, J. R. (1980). Determinants of relapse: Implications for the maintenance of behavior change. In O. Park & S. H. Davidson (Eds.), Behavioral medicine: Changing health lifestyles (pp. 410 – 452). New York: Brunner/Mazel. McMahon, R. C. (2001). Personality, stress, and social support in cocaine relapse prediction. Journal of Substance Abuse Treatment, 21, 77– 87. Monroe, S. M., Kupfer, D. J., & Frank, E. (1992). Life stress and treatment course of recurrent depression: I. Response during index episode. Journal of Consulting and Clinical Psychology, 60, 718 –724. Monroe, S. M., Roberts, J. E., Kupfer, D. J., & Frank, E. (1996). Life stress and treatment course of recurrent depression: II. Postrecovery associations with attrition, symptom course, and recurrence over 3 years. Journal of Abnormal Psychology, 105, 313–328. Monroe, S. M., Slavich, G. M., Torres, L. D., & Gotlib, I. H. (2007). Major life events and major chronic difficulties are differentially associated with history of major depressive episodes. Journal of Abnormal Psychology, 116, 116 –124. Monti, P. M., Colby, S. M., Barnett, N. P., Spirito, A., Rohsenow, D. J., Myers, M., et al. (1999). Brief intervention for harm reduction with alcohol-positive older adolescents in a hospital emergency department. Journal of Consulting and Clinical Psychology, 67, 989 –994. Mun˜oz, R. F., Ying, Y., Perez-Stable, E. J., & Miranda, J. (1993). The prevention of depression: Research and practice. Baltimore: Johns Hopkins University Press. National Institute on Alcohol Abuse and Alcoholism. (2005). Harmful interactions: Mixing alcohol with medicines (NIH Publication No. 03– 5329). Rockville, MD: National Institutes of Health. Nowinski, J., Baker, S., & Carroll, K. (1994). Twelve step facilitation therapy manual. Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism. O’Sullivan, K., Rynne, C., Miller, J., O’Sullivan, S., Fitzpatrick, V., Hux, M., et al. (1988). A follow-up study on alcoholics with and without co-existing affective disorder. British Journal of Psychiatry, 152, 813– 819. Project MATCH Research Group. (1998). Matching alcoholism treatments to client heterogeneity: Treatment main effects and matching effects on drinking during treatment. Journal of Studies on Alcohol, 59, 631– 639. Regier, D. A., Farmer, M. E., Rae, D. S., Locke, B. Z., Keith, S. J., Judd, L. L., et al. (1990). Comorbidity of mental disorders with alcohol and other drug abuse. Journal of the American Medical Association, 264, 2511–2518. Robins, L. N., Wing, J., Wittchen, H. U., Helzer, J. E., Babor, T. F., Burke, J., et al. (1988). The Composite International Diagnostic Interview: An epidemiologic instrument suitable for use in conjunction with different diagnostic systems and in different cultures. Archives of General Psychiatry, 45, 1069 –1077. Rychtarik, R. G., Prue, D. M., Rapp, S. R., & King, A. C. (1992). Self-efficacy, aftercare, and relapse in a treatment program for alcoholics. Journal of Studies on Alcohol, 53, 435– 440. Saunders, B., Baily, S., Phillips, M., & Allsop, S. (1993). Women with alcohol problems: Do they relapse for reasons different to their male counterparts? Addiction, 88, 1413–1422. Sellman, J. D., & Joyce, P. R. (1996). Does depression predict relapse in the 6 months following treatment for men with alcohol dependence? Australian and New Zealand Journal of Psychiatry, 30, 573–578.

SELF-EFFICACY, STRESS, AND TIME TO RELAPSE Shen, X., Hackworth, J., McCabe, H., Lovett, L., Aumage, J., O’Neill, J., & Bull, M. (2006). Characteristics of suicide from 1998 –2001 in a metropolitan area. Death Studies, 30, 859 – 871. Shiffman, S. (1989). Conceptual issues in the study of relapse. In M. Gossop (Ed.), Relapse and addictive behavior (pp. 149 –179). New York: Tavistock/Rutledge. Singer, J. D., & Willett, J. B. (1993). It’s about time: Using discrete-time survival analysis to study duration and the timing of events. Journal of Educational Statistics, 18, 155–195. Sitharthan, T., & Kavanagh, D. J. (1990). Role of self-efficacy in predicting outcomes from a programme for controlled drinking. Drug and Alcohol Dependence, 27, 87–94. Sklar, S. M., Annis, H. M., & Turner, N. E. (1997). Development and validation of the Drug-Taking Confidence Questionnaire: A measure of coping self-efficacy. Addictive Behaviors, 22, 655– 670. Smith, A. J., Hodgson, R. J., Bridgeman, K., & Shepherd, J. P. (2003). A randomized controlled trial of a brief intervention after alcohol-related facial injury. Addiction, 98, 43–52. Sobell, L. C., & Sobell, M. B. (1992). Timeline Follow-Back: A technique for assessing self-reported alcohol consumption. In R. Litten & J. Allen (Eds.), Measuring alcohol consumption: Psychosocial and biochemical methods (pp. 41–72). Totowa, NJ: Humana Press. Stephens, R. S., Wertz, J. S., & Roffman, R. A. (1993). Adult marijuana dependence. In J. S. Baer, G. A. Marlatt, & R. J. McMahon (Eds.), Addictive behaviors across the life span: Prevention and policy issues (pp. 202–218). Newbury Park, CA: Sage. Tate, S. R. (2000). Development of a substance relapse model integrating temporally proximal and distal contextual factors. Dissertation Abstracts International: Section B: The Sciences and Engineering, 61, 2784. (UMI No. 9974113).

57

Tate, S. R., Brown, S. A., Unrod, M., & Ramo, D. E. (2004). Context of relapse for substance-dependent adults with and without comorbid psychiatric disorders. Addictive Behaviors, 29, 1707–1724. Tate, S. R., McQuaid, J. R., & Brown, S. A. (2005). Characteristics of life stressors predictive of substance treatment outcomes. Journal of Substance Abuse Treatment, 29, 107–115. Velasquez, M. M., Carbonari, J. P., & DiClemente, C. C. (1999). Psychiatric severity and behavior change in alcoholism: The relation of the transtheoretical model variables to psychiatric distress in dually diagnosed patients. Addictive Behaviors, 24, 481– 496. Vuchinich, R. E., & Tucker, J. A. (1996). Alcoholic relapse, life events, and behavioral theories of choice: A prospective analysis. Experimental and Clinical Psychopharmacology, 4, 19 –28. Willenbring, M. L. (1986). Measurement of depression in alcoholics. Journal of Studies on Alcohol 47, 367–372. Willett, J. B., & Singer, J. D. (1993). Investigating onset, cessation, relapse, and recovery: Why you should, and how you can, use discrete-time survival analysis to examine event occurrence. Journal of Consulting and Clinical Psychology, 61, 952–965. Willett, J. B., Singer, J. D., & Martin, N. C. (1998). The design and analysis of longitudinal studies of development and psychopathology in context: Statistical models and methodological recommendations. Development and Psychopathology, 10, 395– 426. Witkiewitz, K., & Marlatt, G. A. (2004). Relapse prevention for alcohol and drug problems. American Psychologist, 59, 224 –235.

Received September 26, 2006 Revision received May 23, 2007 Accepted May 24, 2007 䡲

E-Mail Notification of Your Latest Issue Online! Would you like to know when the next issue of your favorite APA journal will be available online? This service is now available to you. Sign up at http://notify.apa.org/ and you will be notified by e-mail when issues of interest to you become available!